Seduction by Contract

SEDUCTION BY CONTRACT
SEDUCTION BY
CONTRACT
law, economics,
and psychology in
consumer markets
OREN BAR-GILL
1
3
Great Clarendon Street, Oxford, ox2 6dp,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© O. Bar-Gill, 2012
The moral rights of the author have been asserted
First Edition published in 2012
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Crown copyright material is reproduced under Class Licence
Number C01P0000148 with the permission of OPSI
and the Queen’s Printer for Scotland
British Library Cataloguing in Publication Data
Data available
Library of Congress Cataloging in Publication Data
Library of Congress Control Number: 2012941355
ISBN 978–0–19–966336–1
Printed in Great Britain
CPI Group (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
For Sigal, Noam, and Guy
Nechama and Aharon
Acknowledgements
This book is the culmination of several years of research and writing on

consumer contracts. This work appeared in several prior publications:
• Credit Card Pricing: The CARD Act and Beyond (with Ryan Bubb),
forthcoming in Cornell Law Review.
• Pricing Misperceptions: Explaining Pricing Structure in the Cell Phone
Service Market (with Rebecca Stone), forthcoming in Journal of Empirical
Legal Studies.
• Product Use Information and the Limits of Voluntary Disclosure (with
Oliver Board), 14 American Law and Economics Review 235, (2012).
• Competition and Consumer Protection: A Behavioral-Economics
Account, in Swedish Competition Authority,The Pros and Cons of Consumer
Protection, ch. 2 (2012).
• Informing Consumers about Themselves (with Franco Ferrari), Sympo-
sium: Juxtaposing Autonomy and Paternalism in Private Law, 3 Erasmus
Law Review 93 (2010).
• Mobile Misperceptions (with Rebecca Stone), 23 Harvard Journal of Law
and Technology 49 (2009).
• The Law, Economics and Psychology of Subprime Mortgage Contracts,
94 Cornell Law Review 1073 (2009).
• Making Credit Safer (with Elizabeth Warren), 157 University of Pennsylva-
nia Law Review 1 (2008).
• The Behavioral Economics of Consumer Contracts, in Oren Bar-Gill and
Richard Epstein, Exchange: Consumer Contracts: Behavioral Economics
vs. Neoclassical Economics, 92 Minnesota Law Review 749 (2008).
• Bundling and Consumer Misperception, Symposium: Homo Economi-
cus, Homo Myopicus, and the Law and Economics of Consumer Choice,
73 University of Chicago Law Review 33 (2006).
• Seduction by Plastic, 98 Northwestern University Law Review 1373 (2004).
viii acknowle dgeme nts
The analysis in this book builds upon and extends the research in these
articles.
My thinking and research on consumer contracts has benefited from the
generous comments and suggestions provided by many colleagues and
friends—on drafts of this book or on the aforementioned articles. For their
comments and suggestions, I would like to thank Barry Adler,Yael Aridor
Bar-Ilan, Jennifer Arlen, Adi Ayal, Ian Ayres, Douglas Baird, Jonathan Baker,
Lily Batchelder,Vicki Been, Jean-Pier Benoit, Susan Block-Lieb, Gabriella
Blum, Jonathan Bolton, Paul Calem, Stephen Choi, Marcus Cole, Robert
Cooter, Richard Craswell, Kevin Davis, Rochelle Dreyfus, Einer Elhauge,
Lee Anne Fennell, Chaim Fershtman, Harry First, Eleanor Fox, Jesse Fried,
Barry Friedman, Mark Geistfeld, Clayton Gillette, David Gilo, Ronald
Gilson, Jeffery Gordon, Solomon Greene, Ofer Grosskopf, Michael Grubb,
Assaf Hamdani, Sharon Hannes, Alon Harel, Claire Hill, Samuel Issacharoff,
Raghuram Iyengar, Christine Jolls, Marcel Kahan, Ehud Kamar, Emir
Kamenica, Louis Kaplow, Aron Katz, Avery Katz, Kevin Kordana, Lewis
Kornhauser, Russell Korobkin, Adriaan Lanni, Michael Levine, Daryl
Levinson, Adam Levitin, Ronald Mann, Yoram Margalioth, Florencia
Marotta Wurgler, Martha Minow, Edward Morrison, Anthony Ogus, Eric
Orts, Gideon Parchomovski, Nicola Persico, Richard Pildes, Eric Posner,
Martin Raphan, Luis Rayo, Elizabeth Renuart, Ariel Rubinstein, Daniel
Schwarcz, Alan Schwartz, Steven Shavell, Howard Shelanski, Peter Siegel-
man, Avi Tabbach, Doron Teichman, Richard Thaler, Diane Thompson,
Avishalom Tor, Jing Tsu, Willem van Boom, Philip Weiser, Lauren Willis,
and Eyal Zamir. I would also like to thank workshop participants at Bar-Ilan,
Berkeley, Chicago, Columbia, Fordham, Haifa, Harvard, IDC Hertzelia,
Jerusalem (Hebrew U.), NYU, Penn, Seton Hall, Stanford, Tel-Aviv, Texas,
University of Illinois, USC, Virginia and Yale, as well as conference partici-
pants at several meetings of the ALEA, EALE, ILEA, at the University of
Chicago Law School conference on Homo Economicus and Homo Myopicus,
at the Rotterdam Workshop on Juxtaposing Autonomy and Paternalism in
Private Law, and at the Swedish Competition Authority Conference on
the Pros and Cons of Consumer Protection.
For excellent research assistance, I thank Efrat Assaf, Michael Biondi,
Osnat Dafna, Joseph Eno, Winnie Fung, Paul McLaughlin, Robin Moore,
Tal Niv, Margot Pollans, Benjamin Roin, Michael Schachter, Lee Schindler,
and James Sullivan. The presentation was substantially improved by James
King, Briony Ryles, and the wonderful team at OUP. I owe special thanks
acknowle dgeme nts ix
to my OUP editor, Alex Flach, for believing in this book from the very
beginning.
I gratefully acknowledge the financial support of the Filomen D’Agostino
and Max E. Greenberg Research Fund at NYU School of Law, the John M.
Olin Center for Law and Economics at Harvard Law School, the William F.
Milton Fund of Harvard University, and the Cegla Center for Interdisciplin-
ary Research of the Law in Tel-Aviv University. I thank the Center for
Customer Relationship Management at Duke University for letting
Rebecca Stone and me use its Telecom Dataset in Chapter 4.
This book was written in one of the most nurturing intellectual environ-
ments imaginable—the NewYork University School of Law. I am immensely
grateful to my Dean, Richard Revesz, for creating such an amazing envir-
onment and for providing limitless support and encouragement for my aca-
demic work. I am also grateful to Tel-Aviv University for its hospitality
during my visit there, when part of this book was written.
I owe special thanks to my co-authors, Oliver Board, Ryan Bubb, Franco
Ferrari, Rebecca Stone, and Elizabeth Warren. It has been a privilege to
work with, and learn from, these amazing colleagues. My thinking, as
reflected in this book, builds, in large part, on my co-authored work. I am
especially grateful to Rebecca for allowing me to use material from our
joint work in Chapter 4.
I am grateful to my “nemesis” Richard Epstein. Debating the theory of
consumer contracts with Richard in the Exchange we published in the
Minnesota Law Review, and beyond, has been an intellectual treat.
I would like to thank my teacher and friend, Ariel Porat, for two decades’
worth of advice and support; and for many great comments on parts of this
book.
Last but not least, I am grateful, beyond words, to my mentors and friends,
Lucian Bebchuk and Omri Ben-Shahar. They put me on the path of schol-
arship. With their intellect, integrity, and dedication, they set the example
that I have been striving to follow. Their advice and guidance, throughout
my career, has been invaluable.Their unforgiving comments have made this
book much better than it otherwise would have been. In fact, there would
be no book without Lucian. He first conceived the idea of this book and
made sure the idea turned into reality.
For Sigal, Noam, and Guy
Nechama and Aharon
Acknowledgements
This book is the culmination of several years of research and writing on

consumer contracts. This work appeared in several prior publications:
• Credit Card Pricing: The CARD Act and Beyond (with Ryan Bubb),
forthcoming in Cornell Law Review.
• Pricing Misperceptions: Explaining Pricing Structure in the Cell Phone
Service Market (with Rebecca Stone), forthcoming in Journal of Empirical
Legal Studies.
• Product Use Information and the Limits of Voluntary Disclosure (with
Oliver Board), 14 American Law and Economics Review 235, (2012).
• Competition and Consumer Protection: A Behavioral-Economics
Account, in Swedish Competition Authority,The Pros and Cons of Consumer
Protection, ch. 2 (2012).
• Informing Consumers about Themselves (with Franco Ferrari), Sympo-
sium: Juxtaposing Autonomy and Paternalism in Private Law, 3 Erasmus
Law Review 93 (2010).
• Mobile Misperceptions (with Rebecca Stone), 23 Harvard Journal of Law
and Technology 49 (2009).
• The Law, Economics and Psychology of Subprime Mortgage Contracts,
94 Cornell Law Review 1073 (2009).
• Making Credit Safer (with Elizabeth Warren), 157 University of Pennsylva-
nia Law Review 1 (2008).
• The Behavioral Economics of Consumer Contracts, in Oren Bar-Gill and
Richard Epstein, Exchange: Consumer Contracts: Behavioral Economics
vs. Neoclassical Economics, 92 Minnesota Law Review 749 (2008).
• Bundling and Consumer Misperception, Symposium: Homo Economi-
cus, Homo Myopicus, and the Law and Economics of Consumer Choice,
73 University of Chicago Law Review 33 (2006).
• Seduction by Plastic, 98 Northwestern University Law Review 1373 (2004).
viii ac k nowle dg em e nts
The analysis in this book builds upon and extends the research in these
articles.
My thinking and research on consumer contracts has benefited from the
generous comments and suggestions provided by many colleagues and
friends—on drafts of this book or on the aforementioned articles. For their
comments and suggestions, I would like to thank Barry Adler,Yael Aridor
Bar-Ilan, Jennifer Arlen, Adi Ayal, Ian Ayres, Douglas Baird, Jonathan Baker,
Lily Batchelder,Vicki Been, Jean-Pier Benoit, Susan Block-Lieb, Gabriella
Blum, Jonathan Bolton, Paul Calem, Stephen Choi, Marcus Cole, Robert
Cooter, Richard Craswell, Kevin Davis, Rochelle Dreyfus, Einer Elhauge,
Lee Anne Fennell, Chaim Fershtman, Harry First, Eleanor Fox, Jesse Fried,
Barry Friedman, Mark Geistfeld, Clayton Gillette, David Gilo, Ronald
Gilson, Jeffery Gordon, Solomon Greene, Ofer Grosskopf, Michael Grubb,
Assaf Hamdani, Sharon Hannes, Alon Harel, Claire Hill, Samuel Issacharoff,
Raghuram Iyengar, Christine Jolls, Marcel Kahan, Ehud Kamar, Emir
Kamenica, Louis Kaplow, Aron Katz, Avery Katz, Kevin Kordana, Lewis
Kornhauser, Russell Korobkin, Adriaan Lanni, Michael Levine, Daryl
Levinson, Adam Levitin, Ronald Mann, Yoram Margalioth, Florencia
Marotta Wurgler, Martha Minow, Edward Morrison, Anthony Ogus, Eric
Orts, Gideon Parchomovski, Nicola Persico, Richard Pildes, Eric Posner,
Martin Raphan, Luis Rayo, Elizabeth Renuart, Ariel Rubinstein, Daniel
Schwarcz, Alan Schwartz, Steven Shavell, Howard Shelanski, Peter Siegel-
man, Avi Tabbach, Doron Teichman, Richard Thaler, Diane Thompson,
Avishalom Tor, Jing Tsu, Willem van Boom, Philip Weiser, Lauren Willis,
and Eyal Zamir. I would also like to thank workshop participants at Bar-Ilan,
Berkeley, Chicago, Columbia, Fordham, Haifa, Harvard, IDC Hertzelia,
Jerusalem (Hebrew U.), NYU, Penn, Seton Hall, Stanford, Tel-Aviv, Texas,
University of Illinois, USC, Virginia and Yale, as well as conference partici-
pants at several meetings of the ALEA, EALE, ILEA, at the University of
Chicago Law School conference on Homo Economicus and Homo Myopicus,
at the Rotterdam Workshop on Juxtaposing Autonomy and Paternalism in
Private Law, and at the Swedish Competition Authority Conference on
the Pros and Cons of Consumer Protection.
For excellent research assistance, I thank Efrat Assaf, Michael Biondi,
Osnat Dafna, Joseph Eno, Winnie Fung, Paul McLaughlin, Robin Moore,
Tal Niv, Margot Pollans, Benjamin Roin, Michael Schachter, Lee Schindler,
and James Sullivan. The presentation was substantially improved by James
King, Briony Ryles, and the wonderful team at OUP. I owe special thanks
ac k nowle dg em e nts ix
to my OUP editor, Alex Flach, for believing in this book from the very
beginning.
I gratefully acknowledge the financial support of the Filomen D’Agostino
and Max E. Greenberg Research Fund at NYU School of Law, the John M.
Olin Center for Law and Economics at Harvard Law School, the William F.
Milton Fund of Harvard University, and the Cegla Center for Interdisciplin-
ary Research of the Law in Tel-Aviv University. I thank the Center for
Customer Relationship Management at Duke University for letting
Rebecca Stone and me use its Telecom Dataset in Chapter 4.
This book was written in one of the most nurturing intellectual environ-
ments imaginable—the NewYork University School of Law. I am immensely
grateful to my Dean, Richard Revesz, for creating such an amazing envir-
onment and for providing limitless support and encouragement for my aca-
demic work. I am also grateful to Tel-Aviv University for its hospitality
during my visit there, when part of this book was written.
I owe special thanks to my co-authors, Oliver Board, Ryan Bubb, Franco
Ferrari, Rebecca Stone, and Elizabeth Warren. It has been a privilege to
work with, and learn from, these amazing colleagues. My thinking, as
reflected in this book, builds, in large part, on my co-authored work. I am
especially grateful to Rebecca for allowing me to use material from our
joint work in Chapter 4.
I am grateful to my “nemesis” Richard Epstein. Debating the theory of
consumer contracts with Richard in the Exchange we published in the
Minnesota Law Review, and beyond, has been an intellectual treat.
I would like to thank my teacher and friend, Ariel Porat, for two decades’
worth of advice and support; and for many great comments on parts of this
book.
Last but not least, I am grateful, beyond words, to my mentors and friends,
Lucian Bebchuk and Omri Ben-Shahar. They put me on the path of schol-
arship. With their intellect, integrity, and dedication, they set the example
that I have been striving to follow. Their advice and guidance, throughout
my career, has been invaluable.Their unforgiving comments have made this
book much better than it otherwise would have been. In fact, there would
be no book without Lucian. He first conceived the idea of this book and
made sure the idea turned into reality.
List of Figures and Tables
Figure 1.1. Factors affecting the total price and total benefit 13
Table 1.1. Benefit from paying late 47
Table 1.2. Total benefit from late payments 47
Table 1.1a(1). Perceived benefit from paying late—Case (1) 48
Table 1.1a(2). Perceived benefit from paying late—Case (2) 48
Figure 2.1. Preference reversal in credit card borrowing 85
Figure 2.2. Borrowing in the near versus the more distant future 114
Figure 4.1. Cumulative distribution functions of cell phone usage 214
Table 4.1. Menu of Three-Part Tariffs 218
Table 4.2a. Summary Statistics—Plan 1 219
Table 4.2b. Summary Statistics—Plan 2 219
Table 4.2c. Summary Statistics—Plan 3 219
Table 4.2d. Summary Statistics—Plan 4 219
Table 4.2e. Summary Statistics—Aggregate 219
Table 4.3a. The likelihood of mistakes 220
Table 4.3b. The magnitude of mistakes 221
Figure 4.2. Plan 3 subscribers—likelihood and magnitude of mistakes 222
Abbreviations
CHAPTER 1
APR Annual Percentage Rate
OIRA Office of Information and Regulatory Affairs
TBO total-benefit-of-ownership
TCO total-cost-of-ownership
CHAPTER 2
CARD Credit Card Accountability, Responsibility, and Disclosure
FRB Federal Reserve Board
NSF no sufficient funds
T&E Travel & Entertainment
TILA Truth-in-Lending Act
CHAPTER 3
ARM adjustable-rate mortgages
CBO Congressional Budget Office
CFPB Consumer Financial Protection Bureau
FDIC Federal Deposit Insurance Corporation
FRM fixed-rate mortgage
FTC Federal Trade Commission
GAO Government Accountability Office
HMDA Home Mortgage Disclosure Act
HOEPA Home Ownership and Equity Protection Act
HUD Housing and Urban Development
LTV loan-to-value
OCC Office of the Comptroller of the Currency
OTS Office of Thrift Supervision
PMI private mortgage insurance
RESPA Real Estate Settlement Procedures Act
xvi ab b rev i ati on s
CHAPTER 4
CDMA Code Division Multiple Access
CPUC California Public Utility Commission
EA Economic Areas
ETF early termination fee
FCC Federal Communications Commission
GSM Global System for Mobile
HHI Herfindahl-Hirschman Index
HV High-Variation
LV Low-Variation
TCO total-cost-of-ownership
TDMA time division multiple access
Introduction
W e are all consumers. As consumers we routinely enter into contracts

with providers of goods and services—from credit cards, mortgages,
cell phones, insurance, cable TV, and internet services to household appli-
ances, theater and sports events, health clubs, magazine subscriptions, trans-
portation, and more.
This book is about consumer contracts. It traces design features common
among multiple types of consumer contracts and explores and explains the
forces responsible for these design features. Why, for example, do sellers
design contracts to provide short-term benefits and impose long-term costs?
Why are low introductory prices so common? Why are cell phones given
away for free, so long as the consumer signs a two-year service contract?
Why are the contracts themselves so complex? What’s the rationale
behind creating credit card and mortgage contracts featuring numerous
fees and interest rates calculated via complex formulas? Why do cellular
service contracts use complicated, three-part tariff pricing—a fixed
monthly fee, a number of included minutes, and an overage fee for min-
utes used beyond the plan limit—and then further complicate matters by
distinguishing between peak minutes, night and weekend minutes, in-
network and out-of-network minutes, minutes used to call a pre-set list of
“friends,” and minutes used to call everyone else? Separate and equally
complex pricing structures are developed and enforced for messaging and
data services.
While clearly contributing to a consumer contract’s complexity, a con-
tract’s “fine print” is not the focus of this book. That no one reads the fine
print is old news. That sellers hide one-sided terms in the fine print is not
surprising. The goal of Seduction by Contract is to explain the design of pric-
ing structures and other contract terms that are often clearly disclosed—
“dickered terms” that consumers are aware of and consent to.
Seduction by Contract. Oren Bar-Gill.

© Oxford University Press 2012. Published 2012 by Oxford University Press.
2 se duc t i on by contract
Market Forces and Consumer Psychology

A main theme of this book is that the design of consumer contracts can
be explained as the result of the interaction between market forces and
consumer psychology. We consumers are imperfectly rational, our deci-
sions and choices influenced by bias and misperception. Moreover, the
mistakes we make are systematic and predictable. Sellers respond to those
mistakes. They design products, contracts, and pricing schemes to maxi-
mize not the true (net) benefit from their product, but the (net) benefit
as perceived by the imperfectly rational consumer. Consumers are lured, by
contract design, to purchase products and services that appear more
attractive than they really are. This Seduction by Contract results in a behav-
ioral market failure.
Competition, many believe, works to increase efficiency and protect
consumers. But competition does not alleviate the behavioral market fail-
ure. It may even exacerbate it. Here’s why: In a competitive market, sellers
have no choice but to align contract design with the psychology of con-
sumers. A high-road seller who offers what she knows to be the best con-
tract will lose business to the low-road seller who offers what the consumer
mistakenly believes to be the best contract. Put bluntly, competition forces
sellers to exploit the biases and misperceptions of their customers.
The interaction between market forces and consumer psychology
explains many of the complex design features so common in consumer
contracts.The temporal ordering of costs and benefits—with benefits accel-
erated and costs deferred—is linked to consumer myopia and optimism.
Complexity responds to bounded rationality, to the challenge of remem-
bering and then aggregating multiple dimensions of costs and benefits.
These two features—complexity and cost deferral—serve one ultimate
purpose: to maximize the (net) benefit from the product, as perceived by
the imperfectly rational consumer.
This behavioral-economics theory is subject to the standard critique that
questions the robustness of the underlying biases and misperceptions in a
market setting. I take this critique seriously. Therefore, I begin by studying
possible rational-choice, efficiency-based explanations for specific contract
design features. Only when I conclude that the rational-choice accounts are
unconvincing or incomplete do I develop the behavioral economics alter-
native.The prevalence of contracts and prices that cannot be fully explained
i nt roduc ti on 3
within a rational-choice framework proves the robustness of the biases and

misperceptions driving the behavioral-economics theory.
A main goal of this book is to explain the design of consumer contracts.
But the descriptive story is the beginning, not the end. Understanding con-
sumer contracts as the product of an interaction between market forces and
consumer psychology raises a host of normative questions and legal policy
challenges. Addressing these questions and challenges is another main ambi-
tion of this book.
Social Costs of the Behavioral Market Failure

The behavioral market failure mentioned earlier threatens social welfare at
several levels. As contractual complexity increases in response to consumers’
imperfect rationality, the cost of comparison shopping also increases, result-
ing in hindered competition. Recall the complex, multidimensional cell
phone contract. Now imagine an imperfectly rational consumer trying to
choose among several such complex, multidimensional contracts. The task
is a daunting one. Many consumers will simply avoid it. Markets don’t work
well when consumers do not shop for the best deal. In the absence of effect-
ive comparison-shopping, prices rise, hurting consumers. Also, consumers
who do not compare offers from different sellers might not be matched
with the seller that best fits their needs. This reduces market efficiency.
While hindered competition reduces social welfare, even intense compe-
tition would not ameliorate the behavioral market failure. The competitive
forces that remain would work to maximize the perceived (net) benefit
rather than the actual (net) benefit from the product. Specifically, short-
term, salient prices would be reduced while long-term non-salient prices
would be increased. Examples of this dynamic are the low credit card teaser-
rates accompanied by high back-end fees and rates, or the free cell phones
that you get when you “agree” to pay a host of usage fees and penalties for
the duration of the two-year lock-in contract.
Prices that are salience-based rather than cost-based produce skewed
incentives—for both product choice and product use. And as the perceived
total price falls below the actual total price, demand for the product becomes
artificially inflated. For example and as we will see later in the book, inflated
demand for subprime mortgages, fueled by contract design, contributed to the
subprime expansion and, later, to the subprime meltdown of 2008.
Finally, complexity and deferred costs raise distributional concerns, since

their effects are not evenly distributed across different groups of consumers.
The interaction between market forces and consumer psychology, then,
creates a market failure. This behavioral market failure imposes substantial
welfare costs. Can legal intervention help? A comprehensive answer is
beyond the scope of this book. Instead, the book focuses on a single regula-
tory technique: disclosure mandates. It develops a new approach to disclos-
ure, one that directly responds to the imperfect rationality problem.
Toward More Effective Disclosure Mandates

While existing disclosure mandates focus largely on product attribute infor-
mation—what the product is and what it does—the analysis and findings
described in this book suggest that more attention should be given to the
disclosure of product-use information, namely, how the product will be used
by the consumer. Rational consumers can be expected to predict their
future-use patterns fairly accurately or at least know their own use patterns
better than the sellers do. Disclosure of product-use information by sellers
is, therefore, superfluous in a rational-choice framework.
The same is not true, however, when consumers are imperfectly rational.
As we will see, consumers often have a poor sense of their future use pat-
terns. Do you know how much you talk on your cell phone? How many
text messages you send and receive? How many megabytes of data you use?
Do you how much you will borrow on your credit card? How fast you will
pay down your balance? Whether you will ever require a cash advance?
How likely are you to miss a payment and incur a late fee? The imperfectly
rational consumer can benefit substantially from disclosure of product-use
information—information that sellers, like cell phone companies and credit
card issuers, make it their business to know.
Moreover, the imperfect rationality of consumers suggests that, to be effect-
ive, disclosure regulation must adopt one of the following two strategies:
• First, simple disclosures that target consumers.The idea is to design aggre-
gate, one-dimensional disclosures that facilitate comparison between
competing products. For example, cell phone companies could be required
to disclose the total annual cost of using a cell phone. The disclosure
would combine both product-attribute information and product-use
i nt roduc ti on 5
information. The annual cost of cellular service would combine rate

information with the consumer’s use-pattern information. Such simple,
aggregate disclosures would help imperfectly rational consumers make
better choices.
• Second, re-conceptualized disclosure aimed not at imperfectly rational
consumers, but at sophisticated intermediaries. Accordingly, this disclos-
ure could be more comprehensive and complex. Consider a consumer
who is considering switching from her current cell phone company to a
competing provider. Consider also an intermediary who wishes to help
the consumer identify the best plan for her needs. The intermediary has
information on the different plans offered by all cell phone companies.
The intermediary, however, has little information on how the specific
consumer uses her cell phone. Since different plans can be optimal for
different consumers, depending on their use patterns, not having prod-
uct-use information substantially reduces the ability of the intermediary
to offer the most valuable advice. Now, the consumer’s current cell phone
company has a lot of information on the consumer’s use patterns. It could
be required to disclose this information in electronic form so that the
consumer could forward it to the intermediary. The intermediary could
then combine the product-use information with the information it has
on different plans and provide the consumer with valuable advice.
***
This book aims to tell a general story about consumer contracts. But each
consumer contract or, more accurately, each class of consumer contracts in
each particular market has its own story. Accordingly, after fleshing out com-
mon themes in Chapter 1, the lion’s share of the book is dedicated to three
case studies of three important consumer markets—credit cards (Chapter 2),
mortgages (Chapter 3), and cell phones (Chapter 4).A detailed market-specific
analysis is necessary to fully answer the descriptive question—why do these
contracts look the way they do?—the normative question—what’s wrong
with these contracts?—and the prescriptive question—what can the law do
to help?.
1
The Law, Economics,
and Psychology of Consumer
Contracts
Introduction
Outcomes in consumer markets are the product of interactions between mar-
ket forces and consumer psychology. Most of this book explores these inter-
actions and their legal policy implications in three consumer markets: credit
cards (Chapter 2), mortgages (Chapter 3), and cellular phones (Chapter 4).
These three chapters present case studies that expose the unique features of
economics–psychology interactions in each market. Indeed, they show that
broad generalizations can rarely be drawn, especially when considering if and
how the law should intervene in consumer markets.
While each market is unique, a common methodology can be applied
to analyze different consumer markets. Such a common methodology—a
behavioral-economics methodology—is described in this chapter. The
bulk of the analysis is descriptive, examining how market forces interact
with consumer psychology to produce observed contract design and pric-
ing structures in the market.1 In this chapter, the descriptive analysis begins
with consumer biases and misperceptions and ends with contracts and
1. See Michael S. Barr, Sendhil Mullainathan, and Eldar Shafir, Behaviorally Informed Financial Serv-
ices Regulation (New America Foundation, 2008) (similarly emphasizing the importance of the
interaction between consumer psychology and market forces, but not focusing on the implica-
tions of this interaction for contract design). Important contributions in the burgeoning field of
Behavioral Industrial Organization consider questions of contract design, albeit with less empha-
sis on policy implications. For an excellent recent textbook that summarizes and synthesizes this
important literature—see Ran Spiegler, Bounded Rationality and Industrial Organization (Oxford
University Press, 2011). Spiegler’s text, and the economic theory literature that it summarizes, is
of foundational importance to many of the themes explored in Seduction by Contract. Through-

law, e conom i c s, and psyc h olog y 7
prices. The goal is to highlight predictions of the behavioral-economics

theory. By way of contrast, the case studies presented in the following
chapters begin with the observed contract designs and pricing schemes,
and then seek a theoretical explanation for the observed contracts and
prices. Moreover, the case studies first consider possible rational-choice,
neoclassical economics explanations for the observed contracts and prices.
The behavioral-economics theory enters only after the standard accounts
are shown to be unsatisfactory or incomplete. Indeed, the failure of the
standard approach provides the impetus for developing an alternative,
behavioral-economics theory.
Following the descriptive analysis, the next step is to explore the norma-
tive implications of the outcomes produced by the interaction of market
forces and consumer psychology. Are these contracts and prices enhancing
or diminishing the welfare of the consumer and of society? As noted in the
introduction, when sellers design contracts and prices in response to the
demand generated by imperfectly rational consumers, the result is a behav-
ioral market failure. Several common welfare costs associated with this mar-
ket failure are incurred, including efficiency costs and distributional costs.
These welfare costs are often mitigated by market solutions; specifically,
learning by consumers and education by sellers. Yet these market solutions
are imperfect. Welfare is not maximized. And this opens the door for con-
sidering legal policy interventions. The range of possible policy responses is
broad and it varies from market to market. Here, and in the remainder of the
book, the discussion of policy implications focuses on a single regulatory
tool: disclosure mandates.
I. The Behavioral Economics of Consumer

Contracts
The behavioral-economics theory rests on two tenets:
(1) Consumers’ purchasing and use decisions are affected by systematic
misperceptions
out the chapter, I often cite Spiegler, rather than the original research papers on which he relies.
For an earlier survey of the economics literature—see G. Ellison, “Bounded Rationality in
Industrial Organization,” in R. Blundell,W. K. Newey, and T. Persson (eds.), Advances in Economics
and Econometrics:Theory and Applications, Ninth World Congress,Vol. II, ch. 5 (2006).
(2) Sellers design their products, contracts, and prices in response to these
misperceptions.
That individual decision-making is affected by a myriad of biases and mis-
perceptions is well documented.2 An important question is whether these
biases and misperceptions persist in a market context and are large enough
to influence market outcomes. As we will see in the following chapters, in
many cases the answer to this question is “yes.”
How does consumer psychology influence contract design and pricing
structure? The basic claim is that market forces demand that sellers be atten-
tive to consumer psychology. Sellers who ignore consumer biases and mis-
perceptions will lose business and forfeit revenue and profits. Over time, the
sellers who remain in the market, profitably, will be the ones who have
adapted their contracts and prices to respond, in the most optimal way, to
the psychology of their customers. This general argument is developed in
Section A below. In particular, the interaction between consumer psychol-
ogy and market forces results in two common contract design features:
complexity and cost deferral. Section B describes these features and explains
why they appear in many consumer contracts.
A. Designing Contracts for Biased Consumers

1. General
It is useful to start by reciting the standard, rational-choice framework. This
standard framework will then be adjusted to allow for the introduction of
consumer biases and misperceptions. Juxtaposing the standard and behavioral
frameworks will help compare market outcomes under the two models.3
In the rational-choice framework, a consumer contract provides the con-
sumer with an expected benefit, B, in exchange for an expected price, P. As
elaborated below, both the benefit and the price can be multidimensional.
The number of units sold, which will be referred to as the demand for a
seller’s product, D, is increasing in the benefit that the product provides, B,
and decreasing in the price that the seller charges, P. Demand is a function
2. See, e.g., Daniel Kahneman, Paul Slovic, and Amos Tversky (eds.), Judgment under Uncertainty:
Heuristics and Biases (Cambridge University Press, 1982).
3. A more formal treatment of the issues presented in Sections I.A.1 and I.A.2 is offered in the
Appendix. For an excellent exposition of the economic theory literature on the topics addressed
in Section A—see Spiegler, above note 1. See also M. Armstrong, “Interactions between Com-
petition and Consumer Policy” Competition Policy International 4 (2008) 97.
of benefits and prices: D(B,P). The seller’s revenue is determined by the

number of units sold, namely, the demand for the product multiplied by the
price per unit. The seller’s profit is equal to revenue minus cost.
When consumers are imperfectly rational, suffering from biases and mis-
perceptions, this general framework must be extended as follows: There is a
perceived expected benefit, B̂, which is potentially different from the actual
expected benefit, B. Similarly, there is a perceived expected price, P̂, which
is potentially different from the actual expected price, P. Demand is now a
function of perceived benefits and prices, rather than of actual benefits and
prices: D(B̂,P̂ ). Revenues—and profits—are a function of perceived bene-
fits and prices, which affect the demand for the product, and of the actual
price.
Before proceeding further, the relationship between imperfect informa-
tion and imperfect rationality should be clarified. Rational-choice theory
allows for imperfect information. A divergence between perceived benefits
and prices on the one hand and actual benefits and prices on the other is
also possible in a rational-choice framework with imperfectly informed
consumers. The focus here, however, is on systemic under- and overesti-
mation of benefits and prices. Perfectly rational consumers will not have
systemically biased beliefs; imperfectly rational consumers will. The main
difference is in how perfectly and imperfectly rational consumers deal with
imperfect information. Rational-choice decision-making provides tools for
effectively coping with imperfect information. These tools are not used by
the imperfectly rational consumer. Instead, he relies on heuristics or cogni-
tive rules-of-thumb, which result in predictable, systemic biases and
misperceptions.4
Sellers must keep costs down and revenues up to maximize profits. As we
have seen, revenues are the product of the number of units sold, or the
demand for the product, multiplied by the price per unit. These observa-
tions imply two tradeoffs that determine the seller’s strategy in a rational-
choice framework: First, the seller wants to increase the benefits from the
product in order to increase demand, but increased benefits usually entail
increased costs. The seller will, therefore, increase the benefits only if the
resulting revenue boost more than compensates for the increased costs. The
second tradeoff focuses on the price: a lower price increases demand, but
4. Moreover, while the perfectly rational consumer realizes that she is imperfectly informed, the
imperfectly rational consumer might be blissfully unaware of the extent of his ignorance.
also decreases the revenue per unit sold. The seller will set prices that opti-
mally balance these two effects.5
The tradeoffs that determine a seller’s optimal strategy when facing
rational consumers are muted in the behavioral-economics model with
imperfectly rational consumers. When perceived benefit is different from
actual benefit, a seller may be able to increase demand by raising the per-
ceived benefit without incurring the added cost of raising the actual benefit.
Similarly, when the perceived price is different from the actual price, demand
can be increased by lowering the perceived price, while keeping revenue per
unit up with a high actual price. Sellers benefit from the divergence between
perceived and actual benefits and between perceived and actual prices. They
will design their contracts and prices to maximize this divergence.
2. The Objects of Misperception: Product Attributes and Use Patterns

When consumers are imperfectly rational, demand, D(B̂,P̂ ), increases with
perceived benefit and decreases with perceived price. The problem is that
consumers will overestimate the total benefit and underestimate the total
price, resulting in artificially inflated demand.6 Why would a consumer
overestimate benefits and underestimate prices? To answer this question, we
must identify the factors that determine these benefits and prices. Moreover,
as explained below, identifying the objects of misperception will prove
important to the welfare and policy analysis presented in the latter half of
this chapter.
Benefits and prices are a function of product attributes and use patterns.
Product attributes define what a product is and what it does. They include
product features, contract terms, and prices. For example, the credit limit
(a product feature) and the interest rate (a price term) are attributes of the
credit card product. Product attributes affect the total benefit and total price.
Misperceptions about product attributes lead to misperceptions of the total
benefit and total price.
While product features define what a product is and what it does, use
patterns define how the product is used. A credit card’s borrowing feature,
5. In addition, certain price dimensions affect how the consumer will use the product and thus
the benefit that the consumer derives from the product. These effects also influence the opti-
mal design of products, contracts, and prices, as explained in subsection 2 below.
6. It also possible that consumers will underestimate the total benefit and overestimate the total
price, resulting in artificially deflated demand.This possibility, however, should be less common,
since sellers have strong incentives to alleviate misperceptions that drive consumers away.
for example, is used frequently by some consumers and less frequently by

others. Some borrow heavily, while others not at all. Product use is clearly
affected by the product’s attributes. But, as detailed below, use patterns are
also affected by other factors. Product use affects the total benefit and total
price. Misperceptions about use patterns lead to misperceptions of the total
Given the importance of use patterns, it is worthwhile to conceptualize
the total benefit as a function of the per-use benefit and use level. Similarly,
it is helpful to conceptualize the total price as a function of the per-use
price and the use level.7
The total benefit will be overestimated when a consumer overestimates
the per-use benefit or use level. For example, the total benefit of a credit
card’s borrowing feature will be overestimated if the consumer overesti-
mates the benefit-per-dollar borrowed. The total benefit will also be over-
estimated if the consumer overestimates how much he will borrow.
In this same way, the total price will be underestimated when a consumer
underestimates the per-use price or use level. The total amount paid in
interest will be underestimated if the consumer underestimates the interest
rate, which can be viewed as the per-use price—the price per dollar bor-
rowed. The total amount paid in interest will also be underestimated if the
consumer underestimates how much he will borrow.
Benefit misperception, then, is a function of misperceived per-use bene-
fits or misperceived use levels while price misperception is a function of
misperceived per-use prices or misperceived use levels.
Let’s dig deeper: Why would a consumer misperceive the per-use benefit
or per-use price? Why would a consumer misperceive the intensity with
which he will use the product, or a product feature?
To understand why the per-use benefit might be misperceived, one
must identify the underlying forces affecting the per-use benefit. Misper-
ception of any of these underlying forces will result in misperception of
the per-use benefit. Consider a credit card’s borrowing feature. The benefit
of this feature is a function of the feature itself: Some cards allow for more
borrowing, setting a higher credit limit, while others are more restrictive,
setting a lower credit limit. The per-use benefit from borrowing drops to
7. The benefit per unit of use need not be constant across all use units. Similarly, the price per
unit of use need not be constant across all use units. In some cases, use is a simple binary vari-
able—the product or feature is either used or not.
zero when the credit limit is reached. The benefit from borrowing is also a
function of the consumer’s intertemporal preferences: Some consumers are
more willing than others to finance present consumption with debt. Finally,
the benefit of borrowing is a function of external forces affecting the con-
sumer’s desire or need to borrow, such as present and expected available
income and conditions affecting the demand for funds, for example illness
or unemployment. The per-use benefit from the credit card’s borrowing
feature will be misperceived when the consumer misperceives the credit
limit, his own preferences, or the strength of external forces that create a
desire or need to borrow.
Next, the per-use price.The per-use price is set by the seller. Some prices
are simple, one-dimensional prices. For example, suppose a movie ticket
costs $12. Such a price will rarely be misperceived. Other prices are com-
plex and multidimensional. Think of an adjustable interest rate, subject to
different caps for per-period changes and various rate-increasing triggers. It
is not hard to see how a consumer might come to misperceive the per-use
price—the interest rate applicable to a dollar borrowed.
Finally, use levels.The decision on how often to use a product or product
feature is influenced by the per-use benefit and per-use price. A higher
benefit-per-dollar borrowed (as, for example, when the consumer is between
jobs) will induce more borrowing. A lower interest rate will similarly result
in more borrowing. The use-level choice might also be influenced by
imperfect rationality. For example, consider a credit card feature that allows
the consumer to pay late—a feature that provides potential liquidity benefits
to the rational consumer. This feature will be “used” more often by imper-
fectly rational consumers who forget to pay on time. Misperception of any
of the factors that influence the per-use benefit or the per-use price will
result in misperception of the use level. A failure to recognize one’s imper-
fect rationality will also result in misperception of use levels. In our example,
a consumer who underestimates his forgetfulness will underestimate the
likelihood of “using” the credit card’s late payment feature.
The objects of misperception can now be summarized and categorized.
The first category includes product attributes. These are product features
and per-use prices determined by the seller when designing the product,
contract, and pricing scheme. A consumer makes a product-attribute mis-
take by misperceiving a product feature or a per-use price. As shown above,
product-attribute mistakes can result in misperception of both the total
Consumer
preferences
Per-use Total
External
benefit benefit
forces
Product
feature Use
level
Total
Per-use price
price
Product attributes
Figure 1.1. Factors affecting the total price and total benefit
The second category of objects of misperception includes use levels. Dif-

ferent product features can be used with different degrees of intensity. Con-
sumers often misperceive these use levels. As explained above, how a product
is used is a function of product attributes—product features and per-use
prices—and of other factors that influence the per-use benefit (and also
other factors that influence the use decision, such as forgetfulness). Product-
use mistakes, or use-pattern mistakes, can result in misperception of both
the total benefit and total price.
This analysis is presented graphically in Figure 1.1, which shows how
product attributes affect total benefit and total price, working through the
per-use benefit, the use level, or both. Figure 1.1 also shows the central role
of use levels in determining the total benefit and total price. Since the total
benefit and total price are a function of product attributes and of use levels,
misperceptions of product attributes and use level mistakes result in misper-
ceptions of the total benefit and total price.
The distinction between product attributes and use levels will prove central
to the welfare and policy implications discussed in subsequent sections.
As we’ll see, market forces are less effective in curing product-use mistakes,
suggesting that we should be more concerned about such mistakes. Corres-
pondingly, the policy prescriptions outlined in Part IV emphasize the import-
ance of disclosure mandates that incorporate product-use information.
Despite their importance, product-use information and product-use
mistakes have received little attention. Moreover, at the policy level, dis-
closure regulation has focused largely on product-attribute information.
The implicit assumption seems to have been that use patterns, being a func-
tion of consumer preferences, are known to consumers. But the preceding
analysis showed that product use is also a function of product attributes and
external forces about which consumers might be imperfectly informed.
Moreover, veering away from rational-choice theory, perfect knowledge of
one’s preferences cannot simply be assumed. The central role of use levels
as an object of misperception thus highlights the value of a behavioral-
economics approach.
3. A Simple Example
a. Setup
A consumer obtains a credit card. The consumer uses the card for transact-
ing only, intending to pay the full balance each month. This consumer,
however, is a bit forgetful. Specifically, he will miss the payment due date
exactly one time each year.
The credit card issuer incurs a fixed annual cost of 4, which represents a
general account maintenance cost. The issuer also incurs a variable, or per-
use, cost of 2 for each late payment. This represents the cost of processing a
late payment and the added risk of default implied by a late payment.
The issuer is contemplating a two-dimensional pricing scheme, includ-
ing an annual fee (p1) and a late fee (p2).The total price equals the annual fee
plus the late fee: p1 + p2. More generally, when the number of late payments
can vary—some consumers never pay late, some pay late once, some pay late
twice, and so on—the total price would equal the annual fee plus the late
fee multiplied by the number of late payments. (The late fee is the per-use
price of the late payment feature; the number of late payments is the use
level.)
b. Misperceptions
A sophisticated, perfectly rational, consumer realizes that he will pay late
once a year and incur a late fee.The sophisticated consumer accurately per-
ceives the total price to be p1 + p2. An imperfectly rational, naive consumer,
on the other hand, underestimates his forgetfulness and mistakenly believes
that he will never pay late and never incur a late fee. The total price, as per-
ceived by the naive consumer, is p1. This divergence between the actual and
perceived total price will affect the equilibrium pricing scheme.
c. Contract Design
Now let’s consider how the issuer will design the credit card contract. How
will the magnitudes of the annual fee (p1) and late fee (p2) be determined?
The answer depends on consumer psychology and market structure. For

ease of exposition and to focus attention on the effect of consumer misper-
ception, assume that the issuer is operating in a competitive market and thus
will set prices that just cover the cost of providing the credit card.
Recall that the issuer faces a fixed annual cost of 4 and a variable cost of
2 per incidence of late payment. Accordingly, the efficient contract sets p1 =
4 and p2 = 2. By setting each price equal to the corresponding cost, the (4,2)
contract provides optimal incentives and thus maximizes value. The issuer
will offer the efficient (4,2) contract to the sophisticated consumer who
realizes that there will be one late payment. This contract minimizes the
total price paid by the consumer, 4 + 2 = 6, while still covering the issu-
er’s costs. In fact, in this simple example any set of prices, p1 and p2, such that
p1 + p2 = 6, would similarly cover the seller’s costs and minimize the total
price paid by the consumer. In a more general example, the (4,2) contract
would be uniquely efficient.8
The efficient (4,2) contract will not be offered to a naive consumer—the
consumer who mistakenly believes that he will never make a late payment
or incur a late fee. For such a consumer, the perceived total price under the
(4,2) contract is 4.The efficient (4,2) contract will not be offered in equilib-
rium, because other contracts appear more attractive to the naive consumer,
while still covering the issuer’s costs. Consider the (0,6) contract. With this
contract, the imperfectly rational, naive consumer perceives a total price of
0, while the actual price is 6. The naive consumer will thus prefer the (0,6)
contract over the efficient (4,2) contract.
In this example, the underestimation of forgetfulness results in underest-
imation of use levels. Contract design amplifies the effect of this product-
use mistake on the perceived total price. In other words, contract design
is used to minimize the perceived total price by amplifying the effect of
product-use mistakes.When the perceived likelihood of triggering a certain
price dimension—such as a late fee—goes down, the magnitude of the
corresponding price—the late fee—goes up. Non-salient prices rise, while
8. Such an example is developed in the Appendix. In essence, the efficiency of the (4,2) contract
is a result of the general efficiency of marginal cost pricing. A late fee set equal to the issuer’s
cost from late payment provides optimal incentives for consumers—to pay late only when the
benefit to them of late payment exceeds the cost of late payment to the issuer. This advantage
of the (4,2) contract assumes that some consumers make a deliberate ex post decision whether
to pay late, contrary to the simple example, in the text, where late payment is an inadvertent
consequence of forgetfulness.
salient prices are reduced. A price is non-salient, because consumers think

it will never be triggered or because the issue (for example, the possibility
of paying late) never crosses the consumer’s mind.
4. The Limits of Competition

It is widely believed that competition among sellers ensures efficiency and
maximizes welfare. This belief is manifested, for example, in antitrust law
and its focus on monopolies and cartels. Competition is also supposed to
help consumers by keeping prices low.
The behavioral-economics model emphasizes the limits of competition.
The example studied in the previous subsection assumed perfect competi-
tion. But we saw that, in equilibrium, sellers offered an inefficient contract
and consumer welfare was not maximized. The reason is straightforward:
Competition forces sellers to maximize the perceived (net) consumer bene-
fit. When consumers accurately perceive their benefits, competition will
help consumers. But when consumers are imperfectly rational, competi-
tion will maximize the perceived (net) benefit at the expense of the actual
(net) benefit.
Focusing on price: When consumers are perfectly rational, sellers com-
pete by offering a lower price.When consumers are imperfectly rational, sell-
ers compete by designing pricing schemes that create an appearance of a
lower price. The underlying problem is on the demand side of the market;
imperfectly rational consumers generate biased demand. Competition forces
sellers to cater to this biased demand. The result: A behavioral market failure.
Modern, neoclassical economics recognizes that even perfectly competi-
tive markets can fail, because of externalities and asymmetric information.
Behavioral economics adds a third cause for market failure; misperception
and bias.This behavioral market failure is a direct extension of the imperfect
information problem. Rational consumers form unbiased estimates of
imperfectly known values. Faced with similarly limited information, imper-
fectly rational consumers form biased estimates. Unbiased estimates can
cause market failure; biased estimates can cause market failure.
The preceding analysis, and the failure of competition that it predicts,
takes consumers’ biases and misperceptions as exogenously given. But per-
ceptions and misperceptions can be endogenous. In particular, sellers can
influence consumer perceptions. That’s a big part of what marketing is
about. What role does competition play in such cases? On the one hand,
sellers might try to exacerbate biases that increase the perceived benefit and
reduce the perceived price of their products.9 On the other hand, sellers
offering superior, yet underappreciated products and contracts may try to
compete by educating consumers and fighting misperception. (See Section
III.B below for further discussion.)
5. Consumer Heterogeneity
Not all consumers are similarly biased. While some consumers overestimate
the net benefit from a product, others might underestimate it. And some
consumers accurately perceive benefits and prices. How does this heterogen-
eity affect the incidence and severity of behavioral market failures?
Contracts are most likely to be designed in response to consumer psy-
chology when many consumers share a common bias. But sellers will design
contracts in response to consumer misperception even when different con-
sumers suffer from different misperceptions. In some cases, sellers will be
able to identify the different consumer groups and their respective biases
and design different contracts for each group. In other cases, sellers will be
unable to distinguish between the different groups of consumers ex ante.
Still, sellers will be able to offer a menu of contracts. Different contracts in
the menu will be attractive to different groups of consumers, according to
their specific biases and misperceptions.
B. Common Design Features

The interaction between consumer psychology and market forces influences
the design of contracts and pricing schemes. In practice, this influence plays
out most often in two common design features: complexity and deferred
costs. These design features will figure prominently in the three case stud-
ies we’ll examine in later chapters. This section defines and illustrates
complexity and deferred costs as contract design features and explains why
these design features figure prominently in consumer contracts.10
9. See Edward L. Glaeser, “Psychology and the Market”, (2004) 94 Amer. Econ. Rev. Papers &
Proceedings 408, 409–11 (“Markets do not eliminate (and often exacerbate) irrationality”; “The
advertising industry is the most important economic example of these systematic attempts to
mislead, where suppliers attempt to convince buyers that their products will yield remarkable
benefits.” “It is certainly not true that competition ensures that false beliefs will be dissipated.
Indeed in many cases competition will work to increase the supply of these falsehoods.”)
10. Complexity and deferred costs are also recurring themes in theoretical models of industrial
organization with imperfectly rational consumers—see Spiegler above note 1, ch. 12. See also
Armstrong, above note 3.
1. Complexity
Most consumer contracts are complex, offering multidimensional benefits
and charging multidimensional prices. Complexity and multidimensional-
ity can be both efficient and beneficial to consumers. Compare a simple
credit card contract with only an annual fee and a basic interest rate for
purchases to a complex credit card contract that, on top of these two price
dimensions, adds a default interest rate, a late fee, and a cash-advance fee.
The complex card facilitates risk-based pricing and tailoring of optional
services to heterogeneous consumer needs.The default interest rate and the
late fee allow the issuer to increase the price for riskier consumers. Such
efficient risk-based pricing is impossible with the simple card. Instead, the
single interest rate implies cross-subsidization of high-risk consumers by
low-risk consumers. Similarly, the cash-advance fee allows the issuer to
charge separately for cash-advance services, which benefit some consumers
but not others. Such tailoring of optional services is impossible with the
simple card. Instead, a higher annual fee implies cross-subsidization of con-
sumers who use the cash-advance feature by those who do not.
These efficiency benefits explain some of the complexity and multi-
dimensionality found in consumer contracts. But there is another, behavio-
ral explanation. Complexity hides the true cost of the product from the
imperfectly rational consumer. A rational consumer navigates complexity
with ease, assessing the probability of triggering each rate, fee, and penalty,
and then calculating the expected cost associated with each price dimen-
sion. The rational consumer may have imperfect information, but will
nonetheless form unbiased estimates given the information he or she
chooses to collect. Accordingly, each price dimension will be afforded the
appropriate weight in the overall evaluation of the product.
The imperfectly rational consumer, on the other hand, is less capable of
such an accurate assessment. He is unable to calculate prices that are indirectly
specified through complex formulas. Even if he could perform this calcula-
tion, he would be unable to simultaneously consider multiple price dimen-
sions. And even if he could recall all the price dimensions, he would be unable
to calculate the impact of these prices on the total cost of the product. Bottom
line: The imperfectly rational consumer deals with complexity by ignoring it.
He simplifies his decision by overlooking non-salient price dimensions.11 And
11. See Richard H. Thaler, “Mental Accounting Matters”, J. Behav. Decision Making, 12 (1999) 183,
194 (finding that small disaggregated fees are ignored).
he approximates, rather than calculates, the impact of the salient dimensions

that cannot be ignored.
In particular, consumers with limited attention and limited memory
exclude certain price dimensions from consideration. In addition, limited
processing ability prevents consumers from accurately aggregating the dif-
ferent price components into a single, total expected price that would
serve as the basis for choosing the optimal product. So while complexity
may not faze the rational consumer, it will most likely mislead the imper-
fectly rational consumer.
As explained above, sellers design contracts in response to systematic
biases and misperceptions of imperfectly rational consumers. In particular,
they reduce the total price, as perceived by consumers, by decreasing salient
prices and increasing non-salient prices. This strategy depends on the exist-
ence of non-salient prices. In a simple contract, the one or two price dimen-
sions will generally be salient. Only a complex contract will have both
salient and non-salient price dimensions. Complexity thus serves as a tool
for reducing the perceived total price.
Moreover, complexity will increase over time as consumers learn to
incorporate more price dimensions into their decision. If sellers signifi-
cantly increase the magnitude of a non-salient price dimension, consumers
will eventually learn to focus on this price dimension, making it salient.
Sellers then will have to find another non-salient price dimension. When
they run out of non-salient prices in the existing contractual design, they
may create new ones by adding more interest rates, fees, or penalties.12
Another common tactic for increasing complexity is bundling. An exam-
ple is the bundling of handsets and cellular service, cemented with lock-in
contracts and early termination fees. This bundle naturally includes multi-
dimensional pricing; specifically, the price of the handset and the pricing
scheme for cellular service (which is itself rather complex). Arguably, one
of the main goals of bundling is to create a low perceived total price by
12. A series of recent papers in industrial organization argue that firms introduce spurious com-
plexity into tariff structures and by doing so inhibit competition and reduce welfare. See, e.g.,
Glenn Ellison, “A Model of Add-On Pricing”, Q.J. Econ., 120 (2005), 585; Xavier Gabaix and
David Laibson, “Shrouded Attributes, Consumer Myopia, and Information Suppression in
Competitive Markets”, Q.J. Econ., 121 (2006), 505; Ran Spiegler, “Competition over Agents
with Boundedly Rational Expectations”, Theoretical Econ., 1 (2006), 207; Glenn Ellison
and Sara Fisher Ellison, “Search, Obfuscation, and Price Elasticities on the Internet”, Econo-
metrica 77 (2009), 427.
reducing the salient handset price and increasing the non-salient pricing of
cellular service. This pricing strategy could not work without bundling:
Carriers need the high service prices to compensate for the low, below-cost
handset prices. Without bundling, consumers would buy the low-price
handset from one seller and get service from a competing carrier, and the
practice of offering below-cost handset prices would soon come to an end.
There are many other examples of bundling: The credit card bundles trans-
acting and borrowing services. Subprime mortgage contracts bundle secured
credit with inspection, appraisal, and insurance services.
In addition to increasing complexity, bundling is used to create deferred-
cost contract designs. This occurs when a product or service with short-
term benefits and prices is bundled with a product or service with long-term
benefits and prices. In the handsets and cellular service example, bundling
both increases complexity and facilitates cost deferral.13
The focus, thus far, has been on intra-contract complexity. A broader
perspective reveals that a consumer’s decision process is far more compli-
cated than sorting through the complexity of any single contract. Consum-
ers today must choose from among many complex products offered by
competing sellers. At first glance, more choice is better. Consumers are het-
erogeneous, so more products mean that consumers can find the product
that best suits their individual needs. But there’s a catch: Searching for the
right product among a complex maze of complex products is costly, even
for rational consumers. Imperfect rationality exacerbates this cost and may
even discourage consumers from searching altogether. More choice comes at
the expense of meaningful choice.
A clarification is in order: In theory, an incomplete understanding of
complex contracts is consistent with rational-choice theory. Facing a com-
plex contract, a rational consumer would have to spend time reading the
contract and deciphering its meaning. If the cost of attaining perfect infor-
mation and perfect understanding of the contract is high, the rational bor-
rower would stop short of this theoretical ideal. Imperfect rationality can be
viewed as yet another cost of attaining more information and better under-
standing. When this cost component is added, the total cost of becoming
informed goes up, and thus the consumer will end up with less information
13. See, generally, Oren Bar-Gill, “Bundling and Consumer Misperception” U. Chi. L. Rev. 73
(2006), 33. Clearly, sellers use bundling and tying not only to increase complexity but also to
defer costs. Specifically, these tactics can be used for anticompetitive reasons.
and a less complete understanding of the contract. Imperfect rationality,

however, is not simply another cost component. Rational consumers who
decide not to invest in reading and deciphering certain contractual provi-
sions will not assume that these provisions are favorable; in fact, they will
recognize that unread provisions will generally be pro-seller. In contrast,
imperfectly rational consumers will completely ignore the unread or for-
gotten terms or naively assume that they are favorable. Accordingly, a com-
plex, unread term or a hidden fee would lead an imperfectly rational
consumer—but not a rational consumer—to underestimate the total cost of
the product. As a result, the incentive to increase complexity and hide fees
will be stronger in a market with imperfectly rational consumers. The
behavioral-economics theory of contract design is an imperfect-rationality
theory, not an imperfect-information theory.
2. Deferred Costs
Non-salient price dimensions and prices that impose underestimated costs
create opportunities for sellers to reduce the perceived total price of their
product. What makes a price non-salient? What leads consumers to under-
estimate the cost associated with a certain price dimension? While there are
no simple answers to these questions, there is one factor that exerts substan-
tial influence on both salience and perception; time.
The basic claim is that, in many cases, non-contingent, short-run costs
are accurately perceived, while contingent, long-run costs are underesti-
mated. An annual fee is a non-contingent price that needs to be paid, by a
specific date in the near future, regardless of how the consumer uses the
card. This cost will figure prominently into the consumer’s selection from
among competing cards. A late fee, on the other hand, is a contingent price
that will be paid in the more distant and unspecified future only if the con-
sumer makes a late payment.This cost, as we have seen, will often be under-
estimated by the consumer and is, therefore, less likely to affect card choice.
If costs in the present are accurately perceived and future costs are underesti-
mated, market forces will produce deferred-cost contracts.
The importance of the temporal dimension of price and cost can often
be traced back to two underlying forces; myopia and optimism. Myopic
consumers care more about the present and not enough about the future.
It is rational to discount future costs and benefits by the probability that
they will never materialize. It is also rational to consider the time-value of
money; a tax or price deferred is a tax or price saved. It is not rational to
discount the future simply because it is in the future or because the future
seems less real, or harder to picture. Myopia is excessive discounting.
Myopia is common. People are impatient, preferring immediate benefits
even at the expense of future costs.14 Myopia is attributed to the triumph of
the affective system, which is driven primarily by short-term payoffs, over
the deliberative system, which cares about both short-term and longer-
term payoffs. This understanding of myopia, and of intertemporal choice
more generally, is consistent with findings from neuroscience.15
Future costs are also often underestimated because consumers are opti-
mistic. The prevalence of the optimism bias has been confirmed in multiple
studies.16 Optimistic consumers tend to underestimate the probability of
triggering contingent, future costs. They underestimate the likelihood that
the contingency will materialize. For example, an optimistic cardholder
might underestimate the probability of making a late payment, leading her
to underestimate the importance of the late fee.
Similarly, when mortgage contracts set low introductory interest rates
coupled with high long-term rates, optimism may cause the consumer to
underestimate the importance of the high long-term interest rates. In the-
ory, the high long-term rates can be avoided by exiting the mortgage
contract—by selling the house or refinancing the mortgage—before the
high rates kick in. In practice, the availability of both the sale and refinan-
cing options decreases when real-estate prices fall. A borrower who is opti-
mistic about real-estate prices will overestimate the likelihood of exiting
14. See Ted O’Donoghue and Matthew Rabin, “Doing It Now or Later”, Amer. Econ. Rev., 89
(1999), 103 (“[p]eople are impatient—they like to experience rewards soon and delay costs
until later.”); George F. Loewenstein and Ted O’Donoghue (2004) “Animal Spirits: Affective
and Deliberative Processes in Economic Behavior” (May 4), available at SSRN: <http://ssrn
.com/abstract=539843> (“people are often powerfully motivated to take myopic actions”—
actions that produce immediate benefits at the expense of future costs).
15. See Loewenstein and O’Donoghue, above note 14; Samuel McClure, David Laibson, George
Loewenstein, and Jonathan Cohen, “Separate Neural Systems Value Immediate and Delayed
Monetary Rewards” (2004) Science, 306, 503–7.
16. See, e.g., Neil D. Weinstein, “Unrealistic Optimism about Future Life Events”, J. Personality &
Soc. Psychol. 39 (1980), 806 (describing two studies revealing that people tend to be unrealisti-
cally optimistic about future life events); Ola Svenson, “Are We All Less Risky and More
Skillful than Our Fellow Drivers?”, Acta Psychologica 47 (1981), 143, 143 (describing a study
revealing that the majority of people “regard themselves as more skillful and less risky than the
average driver”). Like many other cognitive biases, optimism is context specific. Further evi-
dence of optimism in specific consumer markets will be presented in the case-study
chapters.
the mortgage contract before the high rates kick in. As a result, such an
optimistic borrower will underestimate the importance of the high long-
term rates.
Sophisticated sellers facing imperfectly rational consumers will seek to
reduce the perceived total price of their products without reducing the actual
total price that consumers pay. When consumers are myopic or optimistic,
this wedge between perceived and actual prices can be achieved by back-
loading costs onto long-term price dimensions. The result: Deferred-cost
contracts.
II. Welfare Implications

Complex contracts that defer costs into the future hurt consumers and reduce
welfare.They hinder competition and distort the remaining, weakened forces
of competition, leading to excessively high prices on more salient price
dimensions and excessively low prices on less salient price dimensions. The
welfare costs of hindered competition and distorted competition are detailed
below. General distributional implications of complexity and deferred costs
are also briefly discussed. Other, efficiency and distributive concerns are mar-
ket specific and will be addressed in the case studies.17
A. Hindered Competition
Excessively complex contracts prevent effective comparison-shopping and
thus inhibit competition. Sellers gain market power, increasing profits at the
expense of consumers. Limited competition also imposes a welfare cost in
the form of inefficient allocation, as consumers are not matched with the
most efficient seller.
For competition to work well, consumers must be able to compare the
benefits and costs of different products and choose the one that provides the
best value, given the consumer’s tastes and needs. But as we have seen, gath-
ering information on competing products is costly, and complexity—of the
product or contract—increases this cost. A rational consumer will collect
information until the expected marginal benefit of more information is
17. The welfare implications studied in the theoretical literature on industrial organization with
imperfectly rational consumers partially overlap with the welfare implications discussed in this
Part. See Spiegler, above note 1.
outweighed by the marginal cost of collecting more information.When the

cost of collecting information goes up, the rational consumer will collect
less information. Less information implies weaker competition.
Imperfect rationality exacerbates this problem.The cost of collecting and
processing information is higher for the imperfectly rational consumer.
Moreover, the imperfectly rational consumer might not optimally weigh
the benefits and costs of additional information. Confronted with a com-
plex array of complex products, the consumer might engage in insufficient
collection of information or even avoid comparison-shopping altogether.
Competition is not a cure-all, but it does provide important benefits. Com-
plexity stands in the way of effective competition.18
B. Distorted Competition
Complexity weakens the forces of competition. But even if sellers vigor-
ously competed for consumers, biases and misperceptions on the demand
side of the market distort these competitive efforts, leading to suboptimal
outcomes for consumers and reducing social welfare. As explained above,
sellers try to maximize the perceived net benefit of their products in the
eyes of consumers. When consumer perceptions are biased, the products,
contracts, and prices that maximize perceived net benefit are different from
those that maximize actual net benefit. The result is distorted contract
design, with excessive complexity and deferred costs.
Focusing on price, sellers facing rational consumers will try to minimize the
total price of their product. Competition would operate on the total-price level.
Imperfectly rational consumers, on the other hand, choose products based on a
few salient price dimensions. Competition will thus focus on those salient prices,
driving them down, while non-salient prices, free from competitive pressure,
increase. And when salience is a function of time—when short-term prices are
salient and long-term prices are not—competition will drive short-term prices
below cost, with sellers recouping losses through high long-term prices.
These distortions entail two types of efficiency costs.The first pertains to
product choice; the second to how the chosen product is used. Let’s exam-
ine the latter first.
18. For a discussion of additional welfare implications of complex contracts—see David Gilo and
Ariel Porat, “The Hidden Roles of Boilerplate and Standard-Form Contracts: Strategic
Imposition of Transaction Costs, Segmentation of Consumers, and Anticompetitive Effects”
Michigan Law Review, 104 (2006), 983.
Prices affect product-use decisions. A high late fee deters late payments.
A low introductory interest rate induces borrowing during the introductory
period. Optimal pricing provides accurate incentives: With an optimal late
fee, consumers will pay late if, and only if, the benefit of paying late out-
weighs the cost of late payment (including the added risk implied by late
payment) to the issuer.With an optimal interest rate, consumers will borrow
if and only if the benefit from borrowing outweighs the issuer’s cost of pro-
viding credit. Optimal pricing tracks the seller’s cost so that consumers pay
the price and use the product only when the benefits to them outweigh
the seller’s cost.This oversimplified account nonetheless offers a sense of the
factors that determine optimal pricing and of the efficiency gains that opti-
mal pricing provides.
When prices are a function of salience rather than cost, efficiency bene-
fits are compromised. Low salient prices will lead to excessive use, high
non-salient prices to insufficient use. Consumers will borrow excessively
during the introductory period and avoid paying late even when the
benefits of paying late exceed the cost to the issuer of a late payment. Dis-
torted competition produces distorted prices, which lead to distorted
incentives.
Back to product choice: Sellers reduce salient prices and increase non-
salient prices in order to minimize the total price as perceived by the imper-
fectly rational consumer. Since the perceived total price will be lower than
the actual total price, biased consumers may well choose a product that costs
more than it is worth to them. The result is inefficient allocation.
This inefficiency exists even with optimal pricing. Here, the non-salient
price dimensions will be ignored or underestimated, reducing the perceived
total price. Distorted contract design exacerbates the problem by backload-
ing more of the total price onto the non-salient, underestimated dimen-
sions. The gap between actual total price and perceived total price
increases—as does the number of consumers who purchase products that
reduce their welfare. Bias and misperception result in artificially inflated
demand. Distorted contract design adds air to the demand balloon.
C. Distributional Concerns
The distributional implications of complexity and deferred costs are, in
large part, market specific.They will be discussed in the case-study chapters.
Still, a few general observations can be made here.
Excessive complexity imposes a larger burden on less sophisticated cus-

tomers and on financially weaker customers who cannot hire advisers to
help them navigate the complexity of products and contracts. Thus, com-
plexity has a regressive distributional effect.
The distributional effect of deferred costs is less clear. Distorted pric-
ing—specifically, low salient or short-term prices and high non-salient or
long-term prices—shift the burden to the group of consumers who are
more likely to pay the high, non-salient prices. In some cases, such as when
the non-salient prices are default or penalty prices (late fees and default
interest rates, for instance), weaker consumers are more likely to shoulder
the burden of the high, non-salient prices. In these cases, the deferred-cost
feature will have a regressive distributional effect.
III. Market Solutions and Their Limits

As we have seen, the interaction of consumer psychology and market forces
can hurt consumers and reduce welfare.The extent of the harm depends, in
large part, on the ability of market forces—specifically, learning by consum-
ers and education by sellers—to address the underlying biases and misper-
ceptions that are responsible for the behavioral market failure.These market
solutions, and their limits, are considered below.19
A. Learning by Consumers
While bias and misperception can lead to mistakes in product choice and
product use, consumers can learn from their own mistakes—intrapersonal
learning—and from the mistakes of others—interpersonal learning—to
avoid repeating these mistakes.20 But how quickly will consumers learn?
The answer is context-dependent. Context affects the efficacy of both
intrapersonal and interpersonal learning.
Starting with intrapersonal learning, the speed with which consumers
learn about a latent product risk will depend on how frequently they use
the product and on how frequently the risk materializes. Credit cards and
19. See also Armstrong, above note 3.

20. See, e.g., Sumit Agarwal, et al., “The Age of Reason: Financial Decisions over the Life-Cycle
and Implications for Regulation” (2009) Brookings Papers on Economic Activity, Issue 2, 51–117
(showing that consumers learn).
cell phones provide more opportunities for this kind of learning than mort-
gages do. Even with credit cards and cell phones, certain risks and costs arise
more frequently than others. For example, finance charges and overage fees
arise more frequently than currency conversion fees and international
roaming charges.21
When low frequency renders intrapersonal learning less effective, inter-
personal learning becomes important. A consumer might take out one or
two mortgage loans in a lifetime. It is therefore less likely that a consumer
will learn about product risks from personal experience. But if a million
consumers take the same type of mortgage loan, it’s likely that some of these
mortgages will default, thereby exposing the different risky design features
in the mortgage contract.
Interpersonal learning, however, is not a panacea and is also context-
dependent. Interpersonal learning is effective, for example, when products
are standardized; less effective when not. With standardized products, when
consumers discover a certain hidden feature of the product, they can share
this information with family and friends. Since the information pertains to
a standardized product, it is relevant to others. But if the product is not
standardized, such interpersonal learning will be less effective.The informa-
tion obtained by one consumer might not be relevant to another consumer
who purchased a different version of the non-standard product.22
When the nature of the product is more broadly defined to include the
potential uses of the product, the group of standardized products shrinks.
The value of a product does not depend only on its intrinsic features, but
also on its potential uses. If different consumers use the product differently,
then an otherwise standardized product becomes functionally non-stand-
ardized. This can inhibit learning: If one consumer uses the product one
way and through use discovers information about the product, there is little
reason to believe that another consumer who uses the product in a different
way will find this information relevant.
21. On the conditions for effective learning and on the limits of learning—see Amos Tversky
and Daniel Kahneman, “Rational Choice and the Framing of Decisions,” in Robin M. Hog-
arth and Melvin W. Reder (eds.), Rational Choice: The Contrast between Economics and Psychol-
ogy (University of Chicago Press, 1987) 90–1. On the limits of learning, even by sophisticated
decisionmakers in “real world” high-stakes environments—see Cade Massey and Richard H.
Thaler, “The Loser’s Curse: Overconfidence vs. Market Efficiency in the National Football
League Draft” (2010), available at SSRN: <http://ssrn.com/abstract=697121> (documenting
persistent bias in NFL draft picks).
22. Even non-standardized products may share standardized features. Interpersonal learning about
these features can be effective.
The distinction between product-attribute information and product-

use information is also relevant for the efficacy of interpersonal learning.
Mistakes about product attributes are more quickly resolved through
interpersonal learning. How quickly these mistakes are resolved depends on
the number of product units or versions that share the attribute and the
extent that the attribute is relevant to different use patterns. For example,
consumers will more quickly learn about high currency conversion fees the
more common these fees are and the higher the number of consumers who
incur the fees while traveling abroad.
Resolving mistakes involving product use is unlikely to happen through
interpersonal learning. Consumers might underestimate how often they
will make a late payment on a credit card or how often they will exceed the
number of allotted minutes on a cell phone calling plan. Even with an other-
wise standardized product, because use patterns vary widely among con-
sumers, interpersonal learning becomes much more difficult.
Reputation represents another, different aspect of interpersonal learning.
Reputation is most often associated with the seller, rather than a product or
contract feature. Therefore, a seller’s reputation is not subject to the non-
standardization problem. But reputation suffers from a different problem: the
information it conveys is less accurate. Sellers with high-quality products and
reliable customer service may enjoy sterling reputations even while adopting
contract designs that maximize profits at the expense of consumers. Of course,
blatant abuses will hurt a seller’s reputation, and sellers will be careful to avoid
such abuses.The concern is with contractual design features that, while harm-
ful, are unlikely to impose a reputational cost. Reputation is even less effective
at the low end of the market. Many consumers seek out sellers with a reputa-
tion for offering lower-quality products at lower prices. These consumers
might not realize, however, that they are getting harmful contract design along
with lower product quality. Having chosen the low-end reputation, sellers will
be undeterred (or less deterred) by the threat of a reputational penalty.
While imperfect, reputation is an important force. It helps minimize the
adverse effects of the behavioral market failure. Since reputation is learned
by consumers and actively nurtured by sellers, it is relevant both to the dis-
cussion of learning in this subsection and to the discussion of education in
the next subsection.
Another form of learning is based on expert advice. Recognizing their
imperfect rationality, consumers take steps to limit the mistakes they make.
They do this by seeking advice and consulting with experts before entering
the market.23 These experts include private consultants and advisers,

government entities that supply information and guidance, and consumer
organizations. Expert advice is clearly useful. But like other forms of learn-
ing, it is not without limits.
First, consumers do not seek advice before each and every purchase or
use decision. When faced with a major decision, consumers are more likely
to take the time and incur the cost of seeking expert advice. They are less
likely to do so when faced with a smaller decision. For example, consumers
are more likely to seek third-party assistance before taking on a substantial
home equity loan. They are less likely to engage in substantial consultations
before deciding to buy a pair of sneakers with their credit card. In many
markets, consumers make a series of small decisions. In these markets, reli-
ance on expert advice is less prevalent.
Second, expert advice, or good expert advice, is not always available.
Certain decisions that consumers face are so complex that even experts
make mistakes. For example, in the mortgage market, available expert advice
on refinancing ignores the option value of postponing the prepayment
decision—an omission that can cost borrowers up to 25 percent of the loan
value.24 Also, advice is useful when the advice-giver has the advice-taker’s
best interest in mind, which is not always the case. Again, the mortgage
market serves as a case in point. Recent legal reform restricted the permis-
sible structure of broker compensation, responding to concerns that mort-
gage brokers were being rewarded for steering the people who relied on
their advice—borrowers—into more expensive loans.25
Finally, expert advice is more helpful in curing product-attribute mis-
takes than in curing product-use mistakes. When the mistake pertains to
individual or idiosyncratic use patterns, experts rarely have the information
needed to correct the mistake.
Rather than relying solely on experts, consumers sometimes rely on
other consumers. Other consumers can facilitate interpersonal learning
about risky products and contracts.They can also help drive out risky prod-
23. Richard A. Epstein, “Second-Order Rationality,” in Edward J. McCaffery and Joel Slemrod
(eds.), Behavioral Public Finance (Russell Sage Foundation, 2006) 355, 361–2.
24. Sumit Agarwal, John C. Driscoll, and David Laibson,“Optimal Mortgage Refinancing: A Closed
Form Solution” 5–6 (2007) (Nat’l Bureau of Econ. Research,Working Paper No. 13487), available at
<http://www.nber.org/papers/w13487>.
25. Federal Reserve Board, Press Release, August 16, 2010, <http://www.federalreserve.gov/
newsevents/press/bcreg/20100816d.htm>. Similar compensation structures continue to cre-
ate conflicts of interest in the insurance market.
ucts and contracts, thus eliminating the need to learn about them.This is the
“informed minority” argument attributed to Alan Schwartz and Louis
Wilde.26 In theory, even a minority of informed, sophisticated consumers
can induce sellers to offer welfare-maximizing products and contracts.
The informed minority argument relies on several assumptions. First, the
informed minority must not be too small—there must be a sufficiently
large number of informed consumers to affect the design of products and
contracts. Second, the preferences of these informed consumers must be
aligned with those of the less sophisticated consumers. Third, sellers must
not be able to segment the market, offering different contracts to sophisti-
cated and less sophisticated borrowers. When these assumptions fail, the
informed minority argument does not apply.27
B. Education by Sellers
Learning is an important demand-side market solution to the problem of
consumer misperception. A different set of market solutions focus on the sup-
ply side: Sellers may invest in correcting consumer misperceptions. Consider
the following, common scenario. Seller A offers a product that is better and
costs more to produce than the product offered by Seller B. Consumers, how-
ever, underestimate the added value from Seller A’s product and thus refuse to
pay the higher price that Seller A charges. In this scenario, Seller A has a
powerful incentive to educate consumers about its product; in other words, to
correct their underestimation of the product’s value (or total net benefit).
But what if several or more sellers offer identical products or different
products that share a certain product risk? Now a collective-action prob-
lem interferes with sellers’ incentives to educate consumers. To illustrate,
let’s say Seller A reduces the product risk and then invests in educating
consumers about the benefits of its superior product. Seller A will attract a
26. See Alan Schwartz and Louis L.Wilde, “Imperfect Information in Markets for Contract Terms:
The Examples of Warranties and Security Interests”, Va. L. Rev., 69 (1983), 1387; Alan Schwartz
and Louis L.Wilde, “Product Quality and Imperfect Information”, Rev. Econ. Stud., 52 (1985),
251.
27. See Oren Bar-Gill and Elizabeth Warren, “Making Credit Safer” (2008) 157 U. Penn. L. Rev. 1,
22–3. For evidence that the number of informed consumers is too small, at least in certain
markets—see Yannis Bakos, Florencia Marotta-Wurgler, and David R. Trossen, “Does Anyone
Read the Fine Print? Testing a Law and Economics Approach to Standard Form Contracts,”
(2009) NYU Law and Economics Research Paper No. 09-40, available at SSRN: <http://ssrn
.com/abstract=1443256>.
lot of business and make a supra-competitive profit. But this is not an equi-
librium. After Seller A invests in consumer education, all the other sellers
will free ride on Seller A’s efforts. They will reduce the product risk, as
Seller A did, and compete away profit that Seller A would have made.
Anticipating such a response, Seller A will realize that the investment in
consumer education may not be recouped. Seller A may, therefore, choose
not to improve the safety of its product; instead, it will continue to offer a
higher-risk product. This collective-action problem can lead to continued
consumer misperception.28
In some markets, the collective-action problem is avoided by a first-
mover advantage enjoyed by Seller A. In other words, if it takes time for
other sellers to copy Seller A’s consumer-friendly product innovation, Seller
A may be able to earn sufficient profits during this time to make the initial
investment in consumer education worthwhile.
Unfortunately, Seller A is unlikely to enjoy a large first-mover advantage
with contract design innovations. To replicate an improvement in a physical
product, competitors need to reconfigure assembly lines. This takes time. To
replicate a contract-design innovation, however, competitors only need to
type and print (or upload on a website) a new contract.
Education by sellers is particularly unreliable when it comes to product-
use information.The forces that potentially drive sellers to disclose product-
attribute information do not apply to product-use information. Disclosing
product-attribute information provides a competitive advantage to the dis-
closing seller, until other sellers are able to copy. This information is dis-
closed to help buyers appreciate the superiority of the disclosing seller’s
product, as compared to competitors’ products. On the other hand, disclos-
ing product-use information might not generate a competitive advantage
for the disclosing seller. As long as the disclosed use patterns are common
to the entire product category, the now-informed consumer may just as
well purchase the product from a non-disclosing seller. Accordingly, sellers
have little reason to voluntarily disclose use-pattern information.29
28. See Howard Beales, Richard Craswell, and Steven Salop, “The Efficient Regulation of
Consumer Information”, J. L. & Econ. 24 (1981), 491, 527 (explaining why sellers might not
disclose both positive and negative information). On the limits of advertising as a mistake-
correction mechanism—see also Xavier Gabaix and David Laibson, “Shrouded Attributes,
Consumer Myopia, and Information Suppression in Competitive Markets”, Q. J. Econ., 121
(2006), 505.
29. See Oren Bar-Gill and Oliver Board, “Product Use Information and the Limits of Voluntary
Disclosure” American Law and Economics Review 14 (2012), 235.
Finally, even apart from the collective-action and product-use informa-

tion problems, sellers might actually prefer not to correct consumer mis-
takes. They might even invest in creating misperception. This manipulation
of consumer perceptions and preferences is, arguably, one of the main goals
of advertising.30
IV. Policy Implications: Disclosure Regulation

The interaction between consumer psychology and market forces cre-
ates a behavioral market failure. This failure reduces welfare and hurts
consumers. Market solutions exist, but their reach is limited. They
mitigate the problem but cannot ameliorate it. The persistence of a mar-
ket failure opens the door for considering the potential role of legal
intervention.
The policy discussion here and in the remainder of the book focuses
on disclosure regulation. This is not because disclosure always works or
because disclosure is always the optimal form of regulatory intervention.
Rather, it is because disclosure mandates are the least intrusive form of
regulation and, thus, the form of regulation most likely to be adopted. It
is also because disclosure mandates, when optimally designed, directly tar-
get the mistakes and misperceptions at the core of the behavioral market
failure.31
The efficacy of disclosure as a regulatory technique for influencing
behavior and improving market outcomes has been recently called into
30. See Glaeser, above note 9. Additional limits on the efficacy of “education by sellers” as a mar-
ket solution are discussed in Spiegler, above note 1, ch. 6.
31. Disclosure mandates are a primary example of soft paternalism (or asymmetric paternalism
or libertarian paternalism)—helping less sophisticated consumers, while imposing minimal
costs on more sophisticated consumers. See Colin Camerer et al., “Regulation for Conserva-
tives: Behavioral Economics and the Case for ‘Asymmetric Paternalism’ ”, U. Penn. L. Rev.,
151 (2003), 1211; Cass R. Sunstein and Richard H. Thaler, “Libertarian Paternalism Is Not
an Oxymoron”, U. Chi. L. Rev., 70 (2003), 1159; Richard H. Thaler and Cass R. Sunstein,
Nudge: Improving Decisions about Health, Wealth and Happiness (Yale University Press, 2008).
On the benefits and limits of disclosure mandates—see also Spiegler, above note 1, chs. 10,
12; Armstrong, above note 3. Other policy tools, specifically education/literacy/numeracy
programs, are synergetic with disclosure regulation, as they increase consumers’ ability to
digest disclosed information, and thus allow for effective disclosure of more, and more com-
prehensive, information.
question by Omri Ben-Shahar and Carl Schneider.32 In the consumer mar-

kets that are the focus of this book, however, evidence suggests that disclos-
ure has been effective to at least some extent.This evidence will be discussed
in the case-study chapters that follow. But even if the efficacy of current
disclosures is limited, this does not rule out the wisdom of using disclosure
as a regulatory strategy. Existing disclosure mandates are poorly designed.
One purpose of this section is to provide regulators with guidance on how
to better design disclosure mandates for consumer markets.
The goal of this section and of the market-specific treatments of disclos-
ure mandates in the following chapters is not to offer a comprehensive
evaluation of existing or proposed regulations. Rather, the goal is to high-
light and elucidate a number of key issues pertaining to optimal disclosure
regulation. Subsection A highlights the importance of disclosing product-
use information. Subsection B discusses optimal design of disclosure man-
dates. Subsection C concludes the section with a look at some normative
concerns raised by disclosure regulation.
Note that the disclosure strategies discussed below, while designed with
the imperfectly rational consumer in mind, can also help rational but imper-
fectly informed consumers.
A. Disclosing Product-Use Information

Section I.A distinguished between product attributes and product use and,
correspondingly, between product-attribute mistakes and product-use mis-
takes. Both types of mistakes interfere with the efficient operation of mar-
kets and hurt consumers. Information can cure mistakes. Disclosing
product-attribute information can reduce product-attribute mistakes; dis-
closing product-use information can reduce product-use mistakes.
Consider credit cards. The interest rate on a credit card and the penalty for
late payment are attributes of the credit card product. Borrowing patterns and
the incidence of late payment indicate how the product is used.The total bene-
fits and costs associated with a product are a function of both product attributes
and use patterns. Total interest paid depends on both the interest rate and the
32. Omri Ben-Shahar and Carl E. Schneider, More Than You Wanted To Know: The Failure of Man-
dated Disclosure (Princeton University Press, forthcoming).
consumer’s evolving balance.Total penalty charges depend on both the late fee
and the frequency of late payment. Consumers who underestimate the interest
rate and late fee—and those who underestimate how much they will borrow
and how often they will pay late—will underestimate the total price of the
credit card product. The important role of information disclosure in reducing
misperception is widely recognized. To a large degree, however, existing and
proposed disclosure mandates focus solely on product-attribute information.33
I believe that disclosure mandates should target product-use information
as well. As emphasized in Section I, consumers need product-use informa-
tion to make optimal decisions. But, as documented in the case-study chap-
ters, consumers do not have access to reliable product-use information and,
as a result, systemically make mistakes about their future use patterns.
Still, the fact that consumers lack product-use information is insufficient,
in and of itself, to justify regulation that mandates disclosure of product-use
information.Two preliminary objections must first be considered: First, dis-
closure only makes sense if sellers have better information than consumers.
While sellers presumptively have better information about the attributes of
the products that they are offering, the opposite presumption is often applied
to use information: consumers are believed to have better information about
how they are going to use the product. In important consumer markets,
however, this presumption is false. The credit card market is such a market.
Duncan McDonald, former general counsel of Citigroup’s Europe and
North America card businesses, noted:
No other industry in the world knows consumers and their transaction
behavior better than the bank card industry. It has turned the analysis of con-
sumers into a science rivaling the studies of DNA. The mathematics of virtu-
ally everything consumers do is stored, updated, categorized, churned, scored,
tested, valued, and compared from every possible angle in hundreds of the
most powerful computers and by among the most creative minds anywhere.
In the past 10 years alone, the transactions of 200 million Americans have been
reviewed in trillions of different ways to minimize bank card risks.34
33. See Oren Bar-Gill and Franco Ferrari, “Informing Consumers about Themselves” Erasmus
Law Review 3 (2010), 93.
34. Duncan A. MacDonald, “Viewpoint: Card Industry Questions Congress Needs to Ask” Amer-
ican Banker, March 23, 2007. See also Charles Duhigg, “What Does Your Credit-Card Com-
pany Know about You?” (2009) New York Times, May 17 (describing the vast amount of
information, especially product-use information, that credit card companies collect, and then
analyze using sophisticated algorithms informed by psychology research).
The cellular service market provides another example. A pricing manager at

a top US cellular service provider commented that “people absolutely think
they know how much they will use [their cell phones] and it’s pretty sur-
prising how wrong they are.”35 Presumably, the pricing manager was com-
paring people’s perceived use patterns to a benchmark of actual use patterns,
known to the provider and its employees. Even in the mortgage market,
lenders often have superior use information on such things as repayment
patterns and the likelihood of default. This will often be statistical or aver-
age-use information, but lenders also have use information on the specific
individual borrower through credit bureaus and information collected over
time as the borrower’s tenure with the lender lengthens.36
Even when sellers have superior use information, disclosure mandates might
not be justified because of the second preliminary objection: Why mandate
disclosure if sellers can be expected to disclose voluntarily? The answer to this
objection is that sellers will not always volunteer the information.This takes us
back to the question about market solutions and their limits. As argued above,
voluntary disclosure, or education by sellers, cannot always be counted upon.
Moreover, product-use information is less likely to be voluntarily disclosed.
The prevalence of rules requiring product-attribute disclosure and the relative
paucity of mandatory product-use disclosure is, in an important sense, exactly
the opposite of what economic theory would recommend.
Product-use disclosures come in two main forms: statistical, average-use
disclosures and individual-use disclosures. Individual-use disclosures, based on
a consumer’s past use patterns, are clearly more effective in supporting optimal
decision-making by consumers. Such disclosures, however, will generally be
feasible only in service markets, such as the credit card and cell phone markets,
where providers have long-term relationships with their customers and collect
use information over the course of these relationships. In other markets, prod-
uct-use disclosures are often limited to statistical information based on the use
patterns of the average consumer, or the average consumer within a certain
demographic. Heterogeneity among consumers limits the value of such aver-
age-use disclosures. Consumer optimism—a “we-are-all-above-average” atti-
tude—also limits the value of average-use disclosures.
35. Michael D., Grubb, “Selling to Overconfident Consumers”, Amer. Econ. Rev. 99 (2009), 1770.
36. Philip, Bond, David K. Musto, and Bilge, Yilmaz, “Predatory Mortgage Lending” (October
10, 2008). FRB of Philadelphia Working Paper No. 08-24. Available at SSRN: <http://ssrn
.com/abstract=1288094>.
Individual-use disclosures, while more effective, are also subject to cer-

tain limits. First, as explained in Section I.A.2, product use is a function of,
among other things, product attributes. Accordingly, when consumers
switch from one product to another product, their use patterns may change.
For example, consider a cell phone user with a 200-minute-per-month
plan—a plan with no per-minute charge for the first 200 minutes and with
substantial overage fees beyond the 200-minute limit—who switches to a
500-minutes-per-month plan. The consumer may start using the phone
more often. Product-use disclosures based on the 200-minute plan may thus
be misleading. Second, individual-use disclosure will inevitably be based on
past use. Past use is an imperfect predictor of future use. Moreover, optimis-
tic consumers with irresponsible past use—frequent late payments, for
example—may think that they’ll do better in the future.
While product-use disclosures are not perfect, consumers will often be bet-
ter off with them than without them. Regulators designing disclosure regimes
should consider seriously the potential role of product-use disclosures.
B. Designing Optimal Disclosure Mandates

Disclosure mandates are prevalent but often ill-conceived. Simply provid-
ing more information will not always help consumers. Heaps of paper
blindly signed at the closing of a mortgage and the impenetrable fine print
of a credit card contract are extreme examples of disclosure regulation
gone wrong. For sophisticated, rational consumers, the cost of reading and
deciphering these complex disclosures often outweighs the benefit. For
imperfectly rational consumers, information overload is an even bigger
problem.37
37. See Richard Craswell, “Taking Information Seriously: Misrepresentation and Nondisclos-
ure in Contract Law and Elsewhere”, Va. L. Rev. 92 (2006), 565 (arguing that provision of
additional information dilutes the effectiveness of existing disclosures); Russell Korobkin,
“Bounded Rationality, Standard Form Contracts, and Unconscionability”, U. Chi. L. Rev.,
70 (2003), 1203 (consumers can process only limited amounts of information); Govern-
ment Accountability Office, “Credit Cards: Increased Complexity in Rates and Fees
Heightens Need for More Effective Disclosures to Consumers” (2006), available at
<http://www.gao.gov/new.items/d06929.pdf> 46 (finding that credit card disclosures
contain too much information); Mark Furletti, “Credit Card Pricing Developments and
Their Disclosure” (2003) Federal Reserve Bank of Philadelphia, Payment Cards Center,
Discussion Paper, 19 available at <http://www.philadelphiafed.org/pcc/papers/2003/
CreditCardPricing_012003.pdf> (concluding that it is not clear that requiring more details
in regulatory disclosures would be useful for consumers).
To be effective, disclosure mandates must adopt one of two general

approaches. The first approach targets consumers directly, as most current
disclosure mandates do. To be effective, however, these disclosures must be
kept simple—not exactly a characteristic of most current disclosure man-
dates. The second approach re-conceptualizes disclosure to target sophisti-
cated intermediaries or sellers rather than consumers directly. These
disclosures can be more comprehensive and more complex.38
1. Simple Disclosures for Consumers

Designing simple, yet useful disclosures that target imperfectly rational con-
sumers is a difficult task. The challenge is dealing with the inherent tension
between providing more information and providing accessible information.
Sellers have a lot of relevant information. A disclosure that is simple enough
for consumers to understand will inevitably exclude some relevant infor-
mation.The goal in designing disclosures, then, should be to maximize what
consumers take away from the disclosure. To do that, regulators must iden-
tify the most important information and present it in the simplest possible
form.
Consumer heterogeneity further complicates matters, since information
that is valuable in the eyes of one consumer may not be as valuable to
another consumer. Regulators need to mandate disclosure of information
that is statistically important; namely important to the majority of consum-
ers. Alternatively, when possible, regulators should mandate disclosure of
individualized information—information tailored to the individual con-
sumer. The minimum payment disclosure mandated by the Credit Card
Accountability, Responsibility and Disclosure Act of 2009 provides an
example. It provides information on the cost of slow repayment, based on
the individual consumer’s credit card balance.39
In many cases, an effective way to provide the most information in the
least complex way is by disclosing total-cost-of-ownership (TCO) infor-
mation. The TCO disclosure is a simple, single-figure disclosure that pro-
vides consumers with an estimate of how much they will ultimately pay for
38. The distinction between simple disclosures for consumers and comprehensive disclosures for
intermediaries and sellers roughly corresponds to the distinction made by OIRA Administra-
tor, Cass R. Sunstein, between summary disclosure and full disclosure. See Cass R. Sunstein,
“Disclosure and Simplification as Regulatory Tools” Memorandum for the Heads of Execu-
tive Departments and Agencies (2010), available at <http://www.whitehouse.gov/sites/
default/files/omb/assets/inforeg/disclosure_principles.pdf>.
39. For more details—see Chapter 2.
the product over the product’s life span. And when the subject of consumer
mistakes is likely to be the product’s benefits, rather than its cost or price, a
total-benefit-of-ownership (TBO) disclosure may be warranted. In some
cases, the disclosed total cost or total benefit can be measured over a speci-
fied period, typically a year. In these cases, the TCO or TBO disclosures
would become annual cost or annual benefit disclosures.40
The TCO and annual cost disclosures and the TBO and annual benefit
disclosures combine product-attribute information with product-use infor-
mation. For example, the annual cost of a cell phone plan would combine
plan rates with information on use patterns to estimate the annual cost of
cellular service. For a new customer, the carrier may have to use average-use
information to calculate the annual cost estimate (unless the consumer
brings individual-use information from previous experience with another
carrier). For existing customers, the carrier should use the individual-use
information that it already has. A similar annual cost disclosure can be
devised for credit cards, again based on use patterns, which include extent
of borrowing, repayment speed, the likelihood of making a late payment,
and so forth. And for mortgages the TCO estimate would account for pos-
sible interest rate hikes, prepayments, and default.
For consumer credit products, and specifically for mortgages, the Annual
Percentage Rate (APR) disclosure was initially envisioned as a type of TCO
disclosure.The APR is based on the finance charge which, in theory, encom-
passes all costs associated with taking the loan. (In practice, many cost
dimensions were excluded from the finance charge definition, as explained
in Chapter 3.) It then takes this total cost, annualizes it, and presents it as a
percentage figure.
These simple, aggregate disclosures, especially the TCO disclosure, are a
direct response to the behavioral market failure identified in the first part of
this chapter. Sellers facing imperfectly rational consumers will design com-
plex, deferred-cost contracts in order to maximize the wedge between the
40. Cf. Sunstein, above note 38, at 5. For an example of a TBO-type disclosure—see the Lifetime
Income Disclosure Act (S. 267; HR. 1534), which would require DC retirement plan admin-
istrators to disclose the stream of guaranteed lifetime annual benefits that a plan participant
could purchase at retirement, given her current retirement savings. Some administrators, e.g.
TIAA-CREF and Vanguard, include such projections in their statements voluntarily. See also
Gopi Shah Goda, Colleen Flaherty Manchester, and Aaron J. Sojourner, “What’s My Account
Really Worth? The Effect of Lifetime Income Disclosure on Retirement Savings” (2011)
RAND Working Paper No. WR-873, available at <http://ssrn.com/abstract=1925064> (show-
ing that these disclosures can be effective).
actual and perceived cost of their products.The TCO disclosure undermines

sellers’ incentives to design such welfare-reducing contracts. Complexity is
used to hide the true cost of the contract by allowing sellers to load costs
onto less salient price dimensions. If sellers are required to provide a TCO
disclosure that aggregates both salient and non-salient prices, complexity
ceases to be a problem for consumers and loses its appeal to sellers.
Similarly, sellers design deferred-cost contracts so that myopic and opti-
mistic consumers will underestimate the cost of the product. A TCO
disclosure that aggregates both short-term and long-term costs into a single
figure that guides consumer choice would substantially reduce sellers’
incentives to defer costs.
TCO and TBO disclosures can help consumers figure out if the benefit
from the product exceeds its cost. But the TCO and TBO disclosures have
another, perhaps more important role: They facilitate competition by pro-
viding a common metric for comparing competing products. This was
clearly one of the main goals of the APR disclosure when it was first intro-
duced in the late 1960s as part of the original Truth-in-Lending Act. Con-
gress did not imagine that the average consumer would understand how the
APR is calculated. But that didn’t matter. All the consumer needed to do
was identify the APRs of the two or more financial products under consid-
eration and choose the product with the lowest APR. Of course, if the APR
is to guide competition in the right direction, it must correspond to the
total cost of the financial product.
The idea of TCO and TBO disclosures as tools for facilitating competi-
tion highlights their relationship with other product rating systems. Product
ratings on certain consumer websites provide an example. Like the TCO
and TBO disclosures, these ratings attempt to aggregate much information
in a simple measure. Like the TBO disclosure, product ratings focus on the
product’s benefits, rather than on its costs. (For many evaluated products, the
cost to the consumer is simply the one-dimensional price tag.)41
In certain markets,TCO and TBO disclosures are not optimal and a multi-
dimensional disclosure may be needed. This is because a single-figure TCO
or TBO disclosure inevitably leaves out relevant information—information
that becomes more critical as consumer heterogeneity increases and when
this heterogeneity cannot be adequately dealt with by incorporating individ-
41. Sanitation ratings for restaurants provide another example of a rating system. Sanitation rat-
ings, however, focus on one aspect of the consumer experience, and are therefore less closely
related to TCO and TBO disclosures.
ual-use information. When designing such disclosures, regulators should be

mindful of the tradeoff between more information and accessible information.
The theory of optimal disclosure design is still not well developed. Most
disclosure mandates are issued without any attempt to devise optimal dis-
closure forms in a scientific manner. In recent years, such regulators as the
Federal Trade Commission, the Federal Reserve Board, and the Consumer
Financial Protection Bureau have begun to employ consumer-testing meth-
ods to identify more effective disclosure forms.42 These efforts should be
extended.
2. Comprehensive Disclosures for Intermediaries and Sellers

The standard disclosure paradigm targets consumers; in other words, the
disclosures are supposed to be read and used by consumers. But disclosures
can also help consumers even when they are not targeted at consumers
directly. Consumers often rely on agents—intermediaries and even sellers—
to help them choose among competing products. These agents, however,
rarely have enough information to effectively advise consumers. Disclosure
regulation can solve this problem.43
Consider a consumer who is at the end of a two-year cellular service
contract. This consumer needs to decide whether to stay with the current
carrier and plan, or switch to a different plan with the same carrier, or
switch to another carrier altogether. The consumer must choose among
many complex products in the search for the optimal cell phone plan, given
his or her particular use patterns. To do that, the consumer could employ
the services of an intermediary like BillShrink.com. The intermediary will
have information on available plans—product-attribute information. But it
will not have information on the consumer’s use patterns. Of course, the
consumer could provide this information, but as suggested earlier (and as
will be demonstrated in Chapter 4), many consumers have a poor sense of
their use patterns.This is where disclosure kicks in.The missing information
exists in the databases of the consumer’s old carrier. Disclosure regulation
could require carriers to provide this information, in electronic form, to the
consumer. The consumer could forward this data to the intermediary, who
will now be in a position to help the consumer choose the product that best
fits the consumer’s use patterns.
42. For more details—see Chapters 2 and 3. See generally Sunstein, above note 38 at 5 (emphasiz-
ing the importance of testing).
43. Cf. Sunstein, above note 38, at 6–7 (discussing full disclosure).
A related model skips the intermediary and relies on competing sellers as

agents of consumers. In the scenario above, for example, the consumer’s
current carrier is at a competitive advantage because it knows the con-
sumer’s use patterns. If the current carrier is required to disclose use infor-
mation in electronic form, the consumer could then forward this information
to competing carriers and ask which of their plans best fits his or her use
patterns. This type of disclosure would level the playing field between the
old carrier and its competitors, to the benefit of the consumer.
This alternative disclosure paradigm avoids the tradeoff between more
information and more accessible information. Since the disclosed infor-
mation is to be used by sophisticated parties—intermediaries or sellers—
rather than directly by consumers, the disclosure can be comprehensive
and complex. Disclosure that benefits consumers without being targeted
directly at consumers has been prominently proposed by Richard Thaler
and Cass Sunstein.44 Sunstein has begun implementing this proposal in
his role as Administrator of the Office of Information and Regulatory
Affairs (OIRA).45 The idea is also beginning to percolate in the relevant
regulatory agencies. For example, the Federal Communications Com-
mission, in a recent Notice of Inquiry, recognized the potential impor-
tance of both electronic disclosure and intermediaries.46 Finally, the
Mydata initiative in the United Kingdom embraces this new disclosure
paradigm.47
C. Normative Questions
When optimally designed, disclosure can facilitate the efficient operation of
market forces and help consumers. This does not mean that disclosure is a
silver bullet. It has both limits and costs. A comprehensive cost-benefit anal-
44. Thaler and Sunstein, above note 31. See also Bar-Gill and Board, above note 29.
45. See Sunstein, above note 38, at 6–7. See also Cass R. Sunstein,“Informing Consumers through
Smart Disclosure” (2011), Memorandum for the Heads of Executive Departments and Agen-
cies, available at <http://www.whitehouse.gov/sites/default/files/omb/inforeg/for-agencies
/informing-consumers-through-smart-disclosure.pdf>.
46. Federal Communications Commission, “Notice of Inquiry: Consumer Information and Dis-
closure”, CG Docket No. 09-158, released August 28, 2009.
47. See Department of Business Innovation and Skills and Cabinet Office Behavioural Insights
Team, Better Choices: Better Deals, April 13, 2011, available at <http://www.cabinetoffice.gov
.uk/resource-library/better-choices-better-deals>.
ysis of disclosure mandates is beyond the scope of this book, but a few nor-
mative questions should be noted.
As a preliminary matter, it is important to recognize that disclosure is not
without cost. Sellers incur costs of collecting, compiling, and distributing
the information. Some of these costs will be borne by consumers as sellers
increase prices to cover the added cost of the disclosure regulation.48 These
costs, however, should not be overestimated. In many consumer markets,
sellers collect and compile the relevant information anyway. Therefore, the
disclosure mandate will only add the relatively minor cost of distributing
the information.
This relates to another, indirect cost of disclosure regulation. If sellers are
required to disclose the information they collect, they will have a weaker
incentive to collect information.49 While this adverse incentive effect is
undeniably true, its magnitude can be expected to be small in many markets
because the business reasons for collecting information will often outweigh
the disclosure disincentive.50 And, if sharing the information with consum-
ers eliminates its value to sellers, then perhaps the information is collected
only to exploit consumers. In that case, a weaker incentive to collect infor-
mation should not raise concern. Moreover, if the information is socially
valuable and there is a concern that it will not be voluntarily collected,
regulation can mandate collection of the information. In fact, disclosure
mandates imply an obligation to collect the information to be disclosed.
Of course, when the information would not have been collected absent the
mandate, the cost of collection constitutes a cost of the disclosure regula-
tion—a cost that will be passed on, at least in part, to consumers.
At the end of the day, the cost of disclosure will need to be compared to
the cost of alternative regulatory techniques and to the cost of no regulatory
intervention—the welfare costs of the behavioral market failure.
Another normative question involves disclosure and paternalism. Disclos-
ure regulation is often praised for being minimally paternalistic, helping
48. Consumers also incur costs of reading the disclosure and processing the disclosed information.
These costs can be minimized by optimal disclosure design, as described above.
49. Cf. Anthony. T. Kronman, “Mistake, Disclosure, Information, and the Law of Contracts”,
J. Legal Stud., 7 (1987), 1 (arguing that contract law disclosure obligations might deter the
acquisition of information).
50. Kronman distinguishes between deliberately acquired information and casually acquired
information, and argues that casually acquired information can be subject to disclosure man-
dates. Id. In Kronman’s terms, much of the information that sellers should disclose is casually
acquired—it would have been acquired by sellers anyway for business reasons.
consumers make better choices, rather than choosing for them. While dis-
closure mandates are indeed less paternalistic than most other forms of reg-
ulation, they are not completely benign or neutral. With simple disclosures,
some information is inevitably lost. Regulators decide what is lost and what
is emphasized. In addition, disclosure mandates can affect market outcomes.
Requiring sellers to prominently disclose a certain product feature can
focus competition on this feature, leading sellers to enhance or improve the
disclosed feature, at the expense of other, non-disclosed features.51 Even
with disclosure, some measure of paternalism cannot be avoided.
Comprehensive disclosure targeting intermediaries or sellers avoids many
of these issues. But this alternative form of disclosure raises its own set of
concerns. The services of intermediaries are costly. Not all consumers will
avail themselves of these services.Weaker consumers are especially less likely
to seek the help of intermediaries. Disclosure mandates that force sellers to
share information—specifically, use information—with competing sellers
do not impose a direct cost on consumers. Still, they require the consumer
to obtain the electronic disclosure and forward it to the competing seller.
Here again, less sophisticated consumers are less likely to take even the rela-
tively simple steps required to benefit from the regulation.
Conclusion
This chapter set out to develop a general approach for analyzing con-
sumer markets and, more specifically, consumer contracts. A behavioral-
economics model was used to explain how consumer psychology interacts
with market forces to influence the design of consumer contracts. The
resulting behavioral market failure entails potentially significant welfare
costs, which market solutions can reduce but not eliminate. Optimally
designed disclosure mandates, while not a panacea, can enhance efficiency
and help consumers.
The goal of this chapter was to outline a general approach and highlight
common themes. But as noted at the outset, general theorizing can take us
only so far. There are many consumer markets and many more consumer
contracts. Each market is embedded in a unique historical, institutional,
political, and legal context. Most importantly, the underlying currents of
51. See also Sunstein, above note 38, at 4.

44 se duc ti on by contract
consumer psychology and market forces, while following common patterns,

manifest in unique ways in different markets.When it comes to considering
regulatory intervention, a detailed market analysis is imperative.The remain-
der of this book will undertake such an analysis in three important con-
sumer markets: the credit card market, the mortgage market, and the cell
phone market.
Appendix
The Appendix contains a more formal presentation of the ideas presented in
Sections I.A.1 and I.A.2. It also provides a more general example, supplementing
and extending the example presented in Section I.A.3.
1. General
In the rational-choice framework, a consumer contract provides the consumer
with a set of benefits (b1, b2, . . .) in exchange for a set of prices ( p1, p2, . . .), while
imposing on the seller a set of costs (c1, c2, . . .). It is useful to think about the
total expected benefit B(b1, b2, . . .), the total expected price P( p1, p2, . . .), and the
total expected cost C(c1, c2, . . .). The number of units sold, which will be referred
to as the demand for a seller’s product, D, is increasing in the benefit that the
product provides, B, and decreasing in the price that the seller charges, P. The
∂D ∂D
demand function is, therefore, D(B, P ), with >0 and <0. The seller’s
∂B ∂P
revenue, R, is given by the number of units sold, i.e., the demand for the product,
multiplied by the price per unit: R(B, P ) = D(B, P ) · P. And the seller’s profit,
Π, is equal to revenue minus cost: Π(B, P, C ) = R(B, P ) − D(B, P ) · C =
D(B, P ) · (P − C ).
When consumers are imperfectly rational, suffering from biases and mispercep-
tions, this general framework must be extended as follows: In addition to the actual
benefits (b1, b2,. . .), there are perceived benefits (b1̂ , b̂2,. . .), which are potentially
different from the actual benefits. And there is a perceived total expected benefit
B̂(b1̂ , b̂2,. . .), which is potentially different from the actual total expected benefit
B(b1, b2,. . .). Similarly, in addition to the actual prices ( p1, p2,. . .), there are perceived
prices (p̂1, p̂2,. . .), which are potentially different from the actual prices. And there is
a perceived total expected price P̂(p̂1, p̂2,. . .), which is potentially different from the
actual total expected price P(p1, p2,. . .). Demand is now a function of perceived
∂D
benefits and prices, rather than of actual benefits and prices: D(B̂, P̂ ), with >0
∂D ∂B̂
and <0. Revenues are a function both of perceived benefits and prices and of
∂P̂
the actual price: R(B̂, P̂, P ) = D(B̂, P̂ ) · P. And so are profits: Π(B̂, P̂, P, C ) = R(B̂,
P̂, P ) − D(B̂, P̂ ) · C = D(B̂, P̂ ) · (P − C ).52
2. The Objects of Misperception

To better understand the objects of misperception, it is useful to introduce the
notion of a product feature, and link the benefits, prices, and costs to different
product features. To fix ideas, think of a credit card with n features: (f1, f2, . . ., fn).
Each feature of the product can be used more or less intensely. Each product
feature can thus be associated with a use intensity level: (l1, l2, . . ., ln). These use
levels can represent both incidence and intensity (as measured in dollars), e.g.,
how many times per year the consumer borrows and how much money is being
borrowed.53
Benefits, prices, and costs are naturally associated with product features and their
use levels. Starting with costs, for each product feature, there is a per-use cost;
namely, a cost for a single incidence of use (or for a single dollar in each incidence
of use), and a total cost, which is a function of the per-use cost and the use level.
Feature i has a per-use cost of ci and a total cost of Ci(ci, li ). The seller’s costs are not
directly relevant to consumers and thus are not an object of (consumer)
misperception.The discussion of costs and their relation to product features and use
levels is provided for completeness.
Prices are objects of consumer misperception. Prices are also commonly associated
with specific product features and their use levels. Prices are often quoted per unit
of use, such that each product feature is associated with a per-use price and a total
price, which is a function of the per-use price and the use level. Feature i has a per-
use price of pi and a total price of Pi( pi, li ).
The benefits from a product can also be linked to the product’s different
features. It is useful to think of a benefit per unit of use and of a total benefit
associated with each product feature. Feature i has a per-use benefit of bi and a
total benefit of Bi(bi, li ).54
52. A divergence between perceived benefits and prices and actual benefits and prices is possible
also in a rational-choice framework, with imperfectly informed consumers. The focus here,
however, is on systemic under- and overestimation of benefits and prices. Perfectly rational
consumers will not have systemically biased beliefs.
53. In some cases, it is useful to add a general feature, capturing general benefits from having the
product, which is unrelated to any use level. More importantly, this allows for a general price
element, which is not a function of use. The example provided in section 3 below includes
such a general feature.
54. The benefit per unit of use need not be constant across all use units. For example, the per-use
benefit associated with a credit card’s late payment feature can be a vector of benefits, including
a benefit from the first late payment, a different benefit from the second late payment, etc.
What determines the per-use benefit? As explained in the chapter, there are
three factors: the product feature itself, the consumer’s own preferences as related
to the feature, and external factors that influence the per-use benefit from the
feature. And what determines the use level? While the intensity with which a
product feature will be used has thus far been taken as given, use levels are
endogenously determined as a function of the per-use benefit and the per-use
price: li = li(bi, pi ). This observation allows for a more subtle characterization of
benefits and prices. The total price of feature i becomes Pi( pi, li(bi, pi )). And the
total benefit from feature i becomes Bi(bi, li(bi, pi )).
With this enriched framework in place, the object of misperception can be more
accurately identified. Benefit misperception will be considered first, followed by price
misperception. The total benefit from the product will be misperceived when the
benefits from one, or more, product features are misperceived. Since the benefit from
a product feature is a function of the per-use benefit and the use level, misperception
of either element will result in misperception of the benefit from the feature. And
since the use level is itself a function of the per-use benefit and the per-use price, the
benefit from a product feature will be misperceived when the consumer misperceives
either the per-use price or one, or more, of the three factors that influence the per-
use benefit—the product feature, consumer preferences, and external forces.
The total price of the product will be misperceived when one or more of
the feature prices is misperceived. The total price of a product feature will be
misperceived when the per-use price or use level is misperceived. And since
the use level is itself a function of the per-use benefit and the per-use price,
the price of a product feature will be misperceived when the consumer
misperceives either the per-use price or one or more of the three factors that
influence the per-use benefit: the product feature, consumer preferences, and
external forces.
3. Example
a. Benefits, Costs, and Prices
Consider a credit card contract with two features: (1) a general feature or features that
may include the convenience of holding the card, access to customer service, etc., and
(2) a late payment feature. A consumer obtains an annual benefit B1 = 7 from the
general feature(s). This benefit is enjoyed by any consumer who holds the card,
independent of any use level. Alternatively, we could define a degenerate use level l1 = 1
and a per-use benefit b1 = 7, which generate a total benefit B1 = l1 · b1 = 7. Occasionally,
the consumer is short on cash and finds it difficult to make the minimum monthly
payment.The consumer, therefore, benefits from the option to pay late—from the late
payment feature. Specifically, there are four instances during the year in which the
consumer could benefit from paying late. The benefits from late payment vary from
one instance to the other, as detailed in Table 1.1.These are per-use benefits.
Table 1.1. Benefit from paying late

Late payment # 1 2 3 4
Benefit 5 5 3 1
When the consumer sees that paying on time is difficult, the decision becomes
whether to make an on-time payment despite this difficulty or to pay late. The
consumer will pay late when the benefit from paying late (which corresponds to
the difficulty of paying on time), as given in Table 1.1, exceeds the late fee charged
by the issuer, as described below. The total benefit to the consumer from late
payments, B2, depends on the per-use benefits and on the use level, i.e., on the
number of instances, in which the consumer decides to pay late (l2); if the consumer
pays late once, then B2(l2 = 1) = 5; if the consumer pays late twice, then B2(l2 = 2)
= 5 + 5 = 10; if the consumer pays late three times, then B2(l2 = 3) = 5 + 5 + 3 =
13, and if the consumer pays late four times, then B2(l2 = 4) = 5 + 5 + 3 + 1 = 14.
The total benefit from late payments, as a function of the number of late payments,
is summarized in Table 1.2.
Table 1.2. Total benefit from late payments

Number of late payments (l2) 1 2 3 4
Total benefit from late payments (B2) 5 10 13 14
The total benefit to the consumer from the credit card is: B(l2) = B1 + B2(l2) = 7
+ B2(l2), where B2(l2) is determined as specified above.
As to costs: The issuer incurs a fixed annual cost of 4—a general account
maintenance cost associated with the general feature(s): C1 = 4. The issuer also
incurs a variable, or per-use, cost of c2 = 2 per incidence of late payment—the cost
of processing a late payment and the added risk of default implied by a late payment.
Total costs associated with late payments are: C2(c2,l2) = l2 · c2 = l2 · 2. The issuer’s
total costs are: C = C1 + C2 = 4 + l2 · c2 = 4 + l2 · 2.
The issuer is contemplating a two-dimensional pricing scheme, including an
annual fee (p1) and a late fee (p2). The annual fee, which can be interpreted as the
price of the general feature(s), is independent of any use level. Or, if we define a
degenerate use level l1 = 1, the annual fee is the per-use price and the total price
associated with the general feature(s) is: P1(p1, l1) = l1 · p1 = 1 · p1 = p1. The late fee
is the per-use price of the late payment feature. The total price of paying late is a
function of this per-use price and of the use level—each year the consumer pays p2
multiplied by the number of late payments per year, l2: P2( p2, l2) = l2 · p2. The total
amount that a consumer pays per year is: P( p1, l1, p2, l2) = P1( p1, l1) + P2( p2, l2) = p1
+ l2 · p2. To simplify notation, the use-level arguments will sometimes be omitted
from the total price functions: P1( p1, l1) = P1( p1) = p1, P2( p2,l2) = P2( p2) = l2 · p2, and
P( p1, l1, p2, l2) = P( p1, p2) = p1 + l2 · p2.
b. Misperceptions
A rational consumer will accurately perceive the benefit, B = B1 + B2, and the
price, P = P1 + P2. An imperfectly rational consumer might not. For the
imperfectly rational consumer, there will be a perceived benefit, B̂ = B̂1 + B̂2,
and a perceived price P̂ = P̂1 + P̂2. The perceived benefits and prices will
generally diverge from the actual benefits and prices. This divergence will affect
the equilibrium pricing scheme.
To see this, the form of consumer (mis)perception needs to be specified. Suppose
the consumer accurately perceives the general benefit from card use, i.e., B̂1=B1= 7
and the amount to be paid in annual fees, i.e., P1̂ =P1=p1. Misperception concerns
the benefits and costs of paying late. There are two cases:
Case (1): The consumer mistakenly thinks that there will never be cash flow problems
or late payments. (Alternatively, the possibility of paying late might never cross the
consumer’s mind.) The misperceived benefits from late payments are listed in Table
1.1a(1) below.
Case (2): The consumer realizes that there will be cash flow problems and thus benefit
from paying late, but underestimates this benefit. Specifically, assume that the con-
sumer underestimates the benefit from paying late in the second and third instances, as
described in Table 1.1a(2) below.
Case (1) and Case (2) are both examples of misperception of per-use benefit(s) that
will result in product-use mistakes.55
Table 1.1a(1). Perceived benefit from paying late—Case (1)

Benefit 5 5 3 1
Perceived benefit 0 0 0 0
Table 1.1a(2). Perceived benefit from paying late—Case (2)

Benefit 5 5 3 1
Perceived benefit 5 1 1 1
55. In this example, the consumer makes a deliberate, and rational, ex post decision whether to
pay late; the imperfect rationality concerns the ex ante misperception about the decision that
will be made ex post. A different example would assume both ex post and ex ante imperfect
rationality: Ex post consumers inadvertently miss the deadline and end up paying late, and ex
ante consumers underestimate the likelihood that they will inadvertently pay late and, conse-
quently, underestimate the total amount they will pay in late fees.
c. Contract Design
How will the issuer design the credit card contract? How will the magnitudes of
the annual fee ( p1) and late fee ( p2) be determined? The answers depend on
consumer psychology and market structure. For ease of exposition and to focus
attention on the effect of consumer misperception, assume that the issuer is
operating in a competitive market and thus will set prices that will just cover the
cost of providing the credit card.
Recall that the issuer faces a fixed annual cost of 4 and a variable cost of 2 per
incidence of late payment. Facing a rational consumer, the issuer will set p1 = 4 and
p2 = 2. The (4,2) contract guarantees that the issuer’s costs are covered. And it
maximizes the net benefit enjoyed by the consumer. With a late fee of 2, the
consumer will make three late payments for a benefit of 13 (see Table 1.2).The total
benefit from the card will be 7 + 13 = 20. The total price will be 4 + 3 · 2 = 10.
And the net benefit will be 10 (= 20 – 10). Any alternative contract will be less
efficient.56
For example, a lower late fee, say 0.5, would induce the consumer to make four
late payments for a total benefit of 14, rather than 13 under the (4,2) contract. But
the resulting increase in the annual fee won’t be worth the extra unit of benefit.
With four late payments, the issuer’s total cost would be 4 + 4 · 2 = 12. Since
revenues from late fees will be only 4 · 0.5 = 2, the issuer will have to charge an
annual fee of 10 to break even. With the alternative (10,0.5) contract, the consumer
will thus face a total price of 10 + 4 · 0.5 = 12 and a net benefit of 7 + 14 − 12 =
9—lower than the net benefit of 10 under the (4,2) contract.
A higher late fee of, say, 4, would also reduce efficiency. With p2 = 4, the consumer
will make two late payments for a total benefit of 10. The issuer’s total cost would
be 4 + 2 · 2 = 8. Since revenues from late fees, 2 · 4 = 8, cover all costs, the issuer,
operating in a competitive market, will set a zero annual fee. With the alternative
(0,4) contract, the consumer will thus face a total price of 0 + 2 · 4 = 8 and a net
benefit of 7 + 10 − 8 = 9—lower than the net benefit of 10 under the (4,2)
contract.
The efficiency of the (4,2) contract is a result of the general efficiency of marginal
cost pricing. A late fee set equal to the issuer’s cost from late payment provides
optimal incentives for consumers—to pay late only when the benefit to them of
late payment exceeds the cost of late payment to the issuer.
The efficient (4,2) contract will not be offered to an imperfectly rational consumer.
Starting with Case (1), the consumer mistakenly believes that there will never be a
benefit realized by paying late and thus that he will never make a late payment and
never incur a late fee. For such a consumer, the perceived benefit from the credit
56. In this example, because of the discrete nature of the benefit function, there are other con-
tracts that are as efficient as the (4,2) contract. In a more general framework, the (4,2) contract
will be strictly more efficient than any other contract.
card is B̂ = B1̂ + B̂2 = 7 + 0 = 7, the perceived total price is P̂(4, 2) = P1̂ (4) + P2̂ (2)
= 4 + 0 = 4, and the perceived net benefit is 3 (= 7 – 4).The efficient (4, 2) contract
will not be offered in equilibrium, because other contracts appear more attractive
to the biased consumer, while still covering the issuer’s costs. Consider the (0,4)
contract, which, as explained above, induces two late payments and just covers the
issuer’s costs given these two late payments. (While ex ante biased consumers believe
that they will never pay late, ex post they will pay late when the benefit exceeds the
late fee.) With this contract, the biased consumer perceives a total benefit of B̂ = B1̂
+ B̂2 = 7 + 0 = 7, a total price of P̂(0, 4) = P̂1(0) + P̂2(4) = 0 + 0 = 0, and a net
benefit of 7 (= 7 – 0). The biased consumer will thus prefer the (0,4) contract over
the efficient (4,2) contract, even though the latter contract provides more value.
Similar results obtain in Case (2), where the consumer recognizes the potential
benefits from paying late, but underestimates these benefits. Specifically, as
described in Table 1.1a(2), the consumer mistakenly thinks that the benefits from
the second and third late payment are 1 each, while in fact they are 5 and 3,
respectively. The efficient (4,2) contract sets a late fee p2 = 2, which means that
the consumer will pay late whenever the benefit of paying late is larger than 2.
Therefore, the consumer will pay late three times, gaining a benefit of 13 (= 5 +
5 + 3) from these late payments, but the consumer mistakenly believes that there
will only be one late payment for a benefit of 5. For this consumer, the perceived
benefit from the credit card is B̂ = B̂1 + B̂2 = 7 + 5 = 12, the perceived total
price is P̂(4, 2) = P̂1(4) + P̂2(2) = 4 + 1 · 2 = 6, and the perceived net benefit is
6 (= 12 – 6). As in Case (1), the efficient(4,2) contract will not be offered in
equilibrium, because other contracts appear more attractive to the biased
consumer, while still covering the issuer’s costs. Consider the (0,4) contract,
which, as in Case (1), induces two late payments and just covers the issuer’s costs
given these two late payments. With this contract, the biased consumer perceives
a total benefit of B̂ = B̂1 + B̂2 = 7 + 5 = 12, a total price of P̂(0, 4) = P̂1(0) + P̂2(4)
= 0 + 1 · 4 = 4, and a net benefit of 8 (= 12 – 4). The biased consumer will thus
prefer the (0, 4) contract over the efficient (4, 2) contract, even though the latter
contract provides more value.
In both cases, while the initial bias pertains to underestimation of benefits,
contract design is used to maximize the underestimation of price, in order to
produce the maximal perceived net benefit. Use patterns provide the link between
perceived benefit and perceived price. Underestimated benefits result in
underestimated use. Contract design amplifies the effect of the product-use mistake
on the perceived total price.
This feature of the example appears in many real-world scenarios. Contract
design is used to minimize the perceived total price, by amplifying the effect of
product-use mistakes. When the perceived likelihood of triggering a certain price
dimension goes down, the magnitude of the corresponding price goes up. Non-
salient prices rise, while salient prices are reduced. A price is non-salient, because
consumers think it will never be triggered, or because the issue (for example, the
possibility of paying late) never crosses the consumer’s mind.
2
Credit Cards
Introduction
Credit cards are a significant socio-economic phenomenon. In 2008, con-
sumers used almost 1.5 billion credit cards—more than 12 cards per
household—to purchase over $2 trillion of goods and services. The average
household spent $18,000 in credit card transactions, which is more than 35
percent of median household income in the U.S. Credit cards are also the
largest source of non-secured credit. Credit card debt amounted to $976
billion in 2008.1
Credit cards have been the subject of heated political debate for many
years.2 Recently, in the wake of the financial crisis, Congress passed the
Credit Card Accountability, Responsibility, and Disclosure (CARD) Act of
2009, which imposes substantial constraints on credit card issuers.The Dodd–
Frank Wall Street Reform and Consumer Protection Act of 2010 established
a new Consumer Financial Protection Bureau charged with more closely
overseeing consumer credit markets, including the credit card market.
1. See Bureau of the Census, Statistical Abstract of the United States: 2011, tbls. 689, 1187 (hereinafter
“2011 Statistical Abstract”) (tbl. 689 lists the number of households: 117,181,000, and median
household income: $50,303; tbl. 1187 lists the total number of credit cards, the total credit card
purchase volume, and the total amount of credit card debt). See also Thomas A. Durkin, “Con-
sumers and Credit Disclosures: Credit Cards and Credit Insurance” (2002) Fed. Reserve Bull.,
April, at 202 (“Much of the growth of consumer credit in recent years has been in the form of
revolving credit, of which credit card credit is the largest component.”); Teresa A. Sullivan,
Elizabeth Warren, and Jay Lawrence Westbrook, The Fragile Middle Class: Americans in Debt (Yale
University Press, 2000) 129 (“As the fastest growing proportion of consumer debt, credit card
debt has led the way to bankruptcy for an increasing number of Americans. . . . ”).
2. In particular, Congress repeatedly debated whether to reinstate usury ceilings. See, e.g., H.R.
78, 100th Cong. (1987), S. 242, 100th Cong. (1987), S. 647, 100th Cong. (1987), H.R. 3769,
102nd Cong. (1991), H.R. 3860, 102nd Cong. (1991), S.AMDTs 1333–34 to S. 543, 102nd Cong.
(1991), H.R. 4132, 103rd Cong. (1994). The many proposed usury bills did not mature into law.
Rather, Congress opted for a mandatory disclosure policy, as part of the Truth-in-Lending Act,
15 U.S.C. §§ 1601 et seq.

This chapter focuses on the credit card contract. It identifies two features
common to most credit card contracts; complexity and deferred costs. In
the pages that follow, I’ll present a behavioral-economics theory of credit
card contracts, and then argue that complexity and deferred costs represent
a strategic response by sophisticated issuers to imperfectly rational cardhold-
ers.There is a behavior market failure in the credit card industry that reduces
efficiency and hurts cardholders. Regulatory intervention can help mini-
mize the adverse effects of this market failure.
A. Contract Design
The common credit card contract is highly complex. The fees and interest
rates are staggering in both number and complexity. There are annual fees,
cash-advance fees, balance-transfer fees, foreign currency-conversion fees,
expedited-payment fees, no-activity fees, late fees, over-limit fees, and
returned-check fees. As for interest rates, in addition to the primary, long-
term interest rate, there are introductory (teaser) rates, rates on balances
transferred from other cards, rates on cash advances, and default interest
rates. That’s not all: Most of these fees and interest rates are variable and
multidimensional, triggered by complex sets of conditions.
Credit card contracts also feature deferred costs and accelerated
benefits. The multiple price dimensions of the credit card do not always
track the issuer’s underlying cost structure. Rather, issuers set lower
prices for salient, usually short-term dimensions and higher prices for
non-salient, usually long-term dimensions. Consider, for example, the
prevalence of low, even zero, introductory interest rates alongside much
higher long-term interest rates—not to mention the even higher default
interest rates. Similarly, we often see no per-transaction fees, and even
negative per-transaction fees positioned as benefits provided by loyalty
programs—alongside high currency-conversion fees, late fees, and over-
limit fees.
B. Contract Design Explained

Why do credit card contracts couple a high level of complexity with cost
deferral? To find out, let’s first explore a series of standard, rational-choice
theories. These theories do a better, though not completely satisfactory, job
of explaining the complexity dimension.
c re di t card s 53
Multiple price dimensions are needed for efficient risk-based pricing and
optimal tailoring of the credit card product to the needs of heterogeneous
cardholders. Deferred costs are more difficult to explain within a rational-
choice, efficiency-based model. Therefore, the design of credit card con-
tracts, especially the deferred-cost feature, cannot be fully explained within
a traditional, rational-choice framework. What’s needed is an alternative,
behavioral-economic theory of the credit card contract.
Complexity plays into the imperfect rationality of cardholders. As we saw
in the previous chapter, the imperfectly rational cardholder deals with com-
plexity by ignoring it. Decision problems are simplified by ignoring non-
salient price dimensions and “guesstimating,” rather than calculating, the
impact of the salient dimensions that cannot be ignored. In particular, the
imperfectly rational cardholder’s limited attention and limited memory
might result in the exclusion of certain price dimensions from consider-
ation. And limited processing ability might prevent cardholders from accu-
rately aggregating the different price components into a single, total expected
price that would serve as the basis for choosing the optimal card.
Increased complexity may be attractive to issuers, as it allows them to
hide the true cost of the credit card in a multidimensional pricing maze. An
issuer who understands the imperfectly rational response to complexity can
leverage complexity to create an appearance of a lower total price without
actually lowering the price. For example, if the currency-conversion fee and
the cash-advance fee are not salient to cardholders, issuers will raise the
magnitude of these price dimensions. Increasing these non-salient prices
will not hurt demand. On the contrary, it will enable the issuer to attract
cardholders by reducing more salient price dimensions or by increasing
more salient benefit dimensions, such as reward points and frequent-flyer
miles. As discussed in the previous chapter, this strategy depends on the
existence of non-salient price dimensions. When the number of price
dimensions goes up, the number of non-salient price dimensions can also be
expected to go up. Issuers thus have a strong incentive to increase complex-
ity and multidimensionality.
The behavioral-economics explanation for deferred costs is based on
evidence that future costs are often underestimated. When future costs
are underestimated, contracts with deferred-cost features become more
attractive to cardholders and thus to issuers. Underestimation of future costs
can be traced back, in large part, to two underlying cognitive biases; myopia
and optimism. Myopic cardholders focus on short-term benefits and pay
insufficient attention to long-term costs. Optimistic cardholders underesti-

mate their future borrowing and, as a result, underestimate the significance
of the many credit card price dimensions that are contingent upon borrow-
ing.There are two main reasons why cardholders might underestimate their
future borrowing: (a) they might underestimate their self-control problems,
and (b) they might underestimate the likelihood of contingencies bearing
economic hardship. Optimism about one’s self-control and about one’s
future financial condition drive the underestimation bias.
When non-salient, long-term dimensions are underestimated or ignored
by cardholders, issuers may raise prices on these dimensions of the credit
card contract. In a competitive market, the high prices on these long-term
dimensions will pay for low prices on the salient, short-term dimensions
that drive demand for the credit card product. This interaction between
consumer psychology and market forces produces contracts with deferred
costs and accelerated benefits.
So far, the credit card contract has been portrayed as a tool designed to
exploit consumer biases. Interestingly, to the extent that the credit card mar-
ket is competitive, issuers must exploit consumers’ imperfect rationality in
order to survive in this market. Issuers that do not take advantage of con-
sumer biases and, instead, offer lower long-term prices with reduced short-
term perks would not succeed in the marketplace. Consumers would fail to
appreciate the value of reduced long-term prices and would take their busi-
ness elsewhere. On the other hand, competition could have a positive effect
by creating incentives for issuers to educate consumers and reduce the biases
and misperceptions that give rise to excessive complexity and cost deferral.
C. Welfare Implications
The imperfect rationality of cardholders and the contractual design that
responds to this imperfect rationality impose several welfare costs. First, the
complexity of credit card contracts makes it difficult for consumers to com-
parison-shop effectively, thus hindering competition in the credit card mar-
ket. Second, since credit card pricing is, in many cases, salience-based rather
than cost-based, these prices provide inefficient incentives for credit card use.
Zero annual and per-transaction fees, coupled with benefits programs, result
in the creation of too many credit cards and excessive use of these cards.
Teaser rates lead to excessive pre-distress borrowing, which renders the
consumer more vulnerable to financial hardships. Moreover, salience-based
c re di t card s 55
pricing creates an appearance of a cheaper product without actually offering

a cheaper product, resulting in artificially inflated demand for credit cards.
Efficiency isn’t the only thing threatened by the distorted pricing in the
credit card market. Since the high borrowing-related costs are paid by finan-
cially weaker cardholders, the behavioral market failure also raises distribu-
tional concerns.
D. Policy Implications
The welfare costs cited above provide a prima facie case for legal intervention.
Cardholder misperceptions that underlie the identified welfare costs also qual-
ify the no-intervention presumption of the freedom-of-contract paradigm. If
a contracting party misconceives the future consequences of the contract, then
the normative power of contractual consent is significantly weakened.
This chapter focuses on one, common regulatory technique; disclosure
mandates. The behavioral-economics model challenges the conventional
disclosure approach, which focuses almost exclusively on disclosure of
product-attribute information such as interest rates and fees. Issuers should
also be required to disclose product-use information. If a consumer mistak-
enly believes that he or she will never make a late payment, disclosing the
magnitude of the late fee will not affect the consumer’s credit card choice.
Because issuers know more about a consumer’s card-related use patterns
than the consumers themselves do, they should be required to share this
information. Issuers should tell consumers how likely they are to pay late.
Legislators and regulators are beginning to recognize the importance of
product-use disclosures. The Dodd–Frank Wall Street Reform and Con-
sumer Protection Act of 2010 imposes a general duty, subject to rules pre-
scribed by the new Consumer Financial Protection Bureau, to disclose
information, including usage data, in markets for consumer financial
products.3 And Federal Reserve Board regulations, which took effect along
with the CARD Act implementing rules, require that issuers disclose, on
the monthly statement, monthly and year-to-date totals of interest charges
and fees—a disclosure that combines product-attribute and product-use
information.4 These are important first steps.
3. Pub. L. 111–203, Title X, Sec. 1033.

4. CFPB, CARD Act Factsheet <http://www.consumerfinance.gov/credit-cards/credit-card-act
/feb2011-factsheet/>.
I conclude this chapter with guidelines for additional steps. At a general

level, the benefits of more information should be balanced against the risk that
information overload will prevent the imperfectly rational consumer from
reading and understanding the disclosed information. There are two solutions
to this problem. The first solution relies on simple, total-cost disclosures that
combine product-attribute and product-use information. Consumers, choos-
ing among different credit card offers, would then face a straightforward task;
choose the card with the lowest total cost.The second solution relies on inter-
mediaries that would digest the information for consumers. Such intermediar-
ies already exist, but they need help from regulators, who could require, for
example, that product-use information be available in electronic form.
The remainder of this chapter is organized as follows:
• Part I provides background on the credit card market.
• Part II describes the common credit card contract, focusing on the two
identified design features: complexity and cost deferral.
• Part III evaluates possible rational-choice, efficiency-based explanations
for the identified design features.
• Part IV develops an alternative, behavioral-economics theory of contract
design in the credit card market.
• Part V identifies the welfare costs of cardholder misperception and of
contracts designed in response to these misperceptions.
• Part VI discusses market solutions, specifically consumer-friendly credit
cards and debit cards.
• Part VII explores legal policy responses, focusing on disclosure regulation.
I. The Credit Card Market
A. Two Functions
What is a credit card? It is a flat, 3 38 " by 2 18 " piece of plastic engraved with
a name and an account number. This “plastic,” as it’s commonly known,
allows its holder to perform two distinct tasks; to transact quickly and effi-
ciently, and to borrow—to finance a purchase, business, or way of life.Though
offered through the same piece of plastic, transacting and financing func-
tions constitute two distinct services provided by the credit card issuer.
c re di t card s 57
Of course, the credit card holder is not required to make use of both
functions. Some transact but do not borrow, using the credit card only as a
method of payment. While the number of people who use a credit card in
this way is not insignificant, most cardholders use both the transacting and
financing services provided by their cards.5
B. A Brief History
The term “credit card” was coined back in 1887 by Edward Bellamy in his
utopian socialist novel, Looking Backward. Bellamy provided a futuristic
account of the year 2000, when credit cards had entirely supplanted cash.
He was not off by much. In 2008, payment cards (credit cards, charge cards,
and debit cards) accounted for more than 50 percent of transaction volume
in the U.S.6
The history of consumer credit in the more modern world began in the
early twentieth century, when Sears Roebuck and Company started lending
money to its customers so that they could buy the Sears products. Thus, the
merchant card (or retail card) was born. Back then, however, the merchant
card was only accepted by the merchant that provided the specific card.7
The path to the contemporary credit card wound its way through the
so-called Travel & Entertainment (T&E) cards, which were special-pur-
pose charge cards that, in time, evolved into all-purpose cards. The Diner’s
Club card led the way in 1949, followed by American Express and Carte
5. See Brian K. Bucks et al., “Changes in U.S. Family Finances from 2004 to 2007: Evidence from
the Survey of Consumer Finances”, Fed. Reserve Bull., vol. 95 (February 2009), A1-A55 (based
on the 2007 Survey of Consumer Finances, 73 percent of families had at least one credit card,
and among those families 60.3 percent carried a balance); David S. Evans and Richard
Schmalensee, Paying with Plastic (MIT Press, 1999) 211 (transactors “comprise roughly a third of
cardholders but account for about half of charge volume”).
6. For Bellamy’s account—see Lawrence M. Ausubel, “The Credit Card Industry: A History, by
Lewis Mandell” J. Econ. Lit. 30 (1992) 1517, 1518 (book review). For actual payment card transac-
tion volume—see 2011 Statistical Abstract, above note 1, tbls. 683, 1186, 1187 (tbl. 683 lists the
number of consumer units: 120,770,000, and the average annual total expenditure per consumer
unit: $50, 486, the product of which is the total annual expenditure: approximately $6.10 trillion;
tbl. 1187 lists the total debit card purchase volume: 1,347 billion; tbl. 1187 lists the total credit card
purchase volume: 2,153 billion; the combined—debit card and credit card—purchase volume is
$3.5 trillion). See also Ronald Mann, “Adopting, Using, and Discarding Paper and Electronic
Payment Instruments: Variation by Age and Race” (2011). FRB of Boston Public Policy Discussion
Paper No. 11–2. Available at SSRN: <http://ssrn.com/abstract=186216>.
7. Sullivan, Warren, and Westbrook, above note 1, at 109 (“Sears, and then other retailers, gave
consumers the credit that banks would not give them.”). The company’s first application form
asked, “How long at your present address?” and “How many cows do you milk?” Id.
Blanche, which entered the market in 1958. The T&E cards, while gradu-
ally evolving into all-purpose cards, remained charge cards, rather than
credit cards, meaning that the balance on these cards was due in full at the
end of each month.8
In the mid-1960s, with the advent of the Visa and MasterCard systems,
the modern credit card was born, combining the all-purpose feature of the
evolved T&E cards with the credit feature of the merchant cards. The Visa
and MasterCard bankcards grew rapidly as they added more and more bank
issuers and merchants to their networks. In 1985, Sears Roebuck and Com-
pany introduced its own all-purpose credit card, the Discover card. Ameri-
can Express joined the credit card scene with its Optima card in the late
1980s. As Sullivan, Warren, and Westbrook put it: “By the 1990s the all-
purpose card gained dominance over traditional store cards, making it pos-
sible for millions of card holders to charge anything from their dental fillings
to their parking tickets.”9
For credit card issuers, the initial steps on the road to success were shaky.
With strict usury laws in place that mandated caps on interest rates, issuers
initially lost money on their credit card business. The banking industry
eventually overcame the strict interest-rate ceilings in an ingenious way.
Instead of lobbying each state legislature for more lenient usury laws, the
industry targeted, in the Federal courts, the jurisdictional issue of the
“exportation” of interest rates, the question being “which state’s usury ceil-
ing constrains the interest rate if a bank located in one state issues a credit
card to a consumer in a different state.”10
In 1978, the exportation question reached the United States Supreme
Court. The Court, in Marquette National Bank v. First of Omaha Service
Corporation, ruled that the applicable usury ceiling was the one set by the
state in which the issuing bank was located. In effect, the Supreme Court
8. Id. See also Lewis Mandell, The Credit Card Industry: A History (Twayne Publishers, 1990) 1–3
(an historical account of the conception of the Diner’s card); Evans and Schmalensee, above
note 5, at 10–11 (describing the entry of the American Express and Carte Blanche cards).
9. Sullivan, Warren, and Westbrook, above note 1, at 109; Evans and Schmalensee, above note 5,
at 10–11.
10. On the shaky initial steps—see Evans and Schmalensee, above note 5, at 68–9, 73; Lawrence
M. Ausubel, “Credit Card Defaults, Credit Card Profits, and Bankruptcy” Am. Bankr. L.J. 71
(1997) 249, 260–1 (“Before 1982, credit card interest rates were subject to usury ceilings in
most states. These ceilings on interest rates limited credit card profitability during periods,
such as 1974–1975 and 1980–1981, when market interest rates on Treasury bills and corporate
bonds spiked upward.This led to a sharply-reduced or negative return on assets for credit card
activity during such years.”) On the interest rate “exportation” solution—see Ausubel, id.
c re di t card s 59
“gave banks the option of shifting their credit card operations to wholly
owned subsidiaries situated in states without usury laws.” The Marquette
decision fired the opening shot in the inter-state race to attract credit card
issuers. To win this race, or to, at minimum, prevent an exodus of banks,
many states substantially increased their interest caps or revoked their usury
laws altogether.Thus, the Marquette decision produced a functionally dereg-
ulated credit card market. Moreover, the decision enabled credit card issuers
to operate on a national level and thus to enjoy economies of scale.11
The sky-high inflation of the late 1970s and early 1980s lifted the final
barrier to the profitability of the credit card industry.With the effective abo-
lition of usury laws, credit card interest rates rose to match the high inflation
rates. In fact, the causation probably worked in both directions; the high
inflation rates were likely instrumental in bringing about the legal rulings
(specifically the Marquette decision) that triggered the abolition of usury ceil-
ings. Whatever the sequence, credit card interest rates rose with inflation.12
As high inflation justified raising interest rates in the late 1970s and early
1980s, the subsequent decline in the inflation rate starting in 1982–83 might
have been expected to bring about a reduction in those rates. This reduc-
tion, however, never came or, more accurately, was very late to arrive (see
below).13
The early history of the credit card industry was marked by declining
costs and “sticky” interest rates. The high interest rates, which stubbornly
failed to keep pace with the declining cost of funds, allowed credit card
issuers to offer more credit and to target less credit-worthy consumers.14
11. The Supreme Court’s decision is at 439 U.S. 299 (1978) (henceforth “the Marquette decision”).
In Marquette the Court interpreted 12 U.S.C. § 85. For the implications of the Marquette
decision—see Lawrence M.Ausubel,“The Failure of Competition in the Credit Card Market”
Amer. Econ. Rev. 81 (1991) 50, 52; Sullivan, Warren, and Westbrook, above note 1, at 248–9;
Evans and Schmalensee, above note 5, at 6, 71–2.
12. See Sullivan, Warren, and Westbrook, above note 1, at 248–9 (“All that changed with the sky-
high inflation of the late 1970s and early 1980s. With inflation in double digits, Congress and
the Supreme Court effectively legalized what had been usury, overriding the restrictive state
laws.”).
13. Sullivan, Warren, and Westbrook, above note 1, at 255 (“[T]he single biggest cost for a credit
card issuer is the cost of funds for the money it lends to borrowers who repay over time.
Between 1980 and 1992, the rate at which banks borrow money fell from 13.4 percent to 3.5
percent. During the same time, the average credit card interest rate rose from 17.3 percent to
17.8 percent.Thus during the period that the credit card issuers’ largest cost was plummeting,
they were raising the price of credit to their consumers.”). See also id. at 18–19, 248–9;
Ausubel, above note 11, at 53–5.
14. Alternatively, the declining cost of funds led issuers to extend credit to less credit-worthy
consumers, and the increased risk prevented the decline in interest rates.
The result: An explosion of consumer credit, which led to a dramatic

expansion of consumer debt and also to an increase in consumer bank-
ruptcy rates.15
The more recent history of the credit card industry, starting in the early
1990s, has been marked by the technological advances that made risk-based
pricing possible.The ability to match interest rates (and other price dimensions)
to borrower risk allowed issuers to serve less creditworthy borrowers, thus
fueling the continuing expansion of credit cards. Risk-based pricing implies
lower interest rates for low-risk consumers and high rates for high-risk con-
sumers. It does not, however, imply a reduction in average rates. But such a
reduction did finally occur, starting in the early 1990s, as competition focused
on the interest rate, which became more salient to consumers.16
C. Economic Significance
Credit cards are a major method of payment, and their prevalence and
importance is only growing. By 1995, credit cards had already surpassed cash
as a method of payment. In 2008, consumers used almost 1.5 billion credit
cards (more than 12 cards per household) to purchase over $2 trillion of
goods and services. The average household completed $18,000 of credit
card transactions, over 35 percent of the median household income.17
The credit card industry has experienced a significant growth rate by any
measure, be it transaction volume, number of cards in circulation, or out-
standing balances. While in 1970 only 16 percent of households had credit
15. Diane Ellis, “The Effect of Consumer Interest Rate Deregulation on Credit Card Volumes,
Charge-Offs and the Personal Bankruptcy Rate” (1998) 98–05 Bank Trends (identifying the
link between the repeal of usury rates, the increase in consumer credit, and the rise in bank-
ruptcy filing rates). The historical link between the rise of the credit card and the explosion
of consumer debt is quite significant. See James Medoff and Andrew Harless, The Indebted
Society: Anatomy of An Ongoing Disaster (Little Brown & Co., 1996) 9 (“Since the introduction
of credit cards, the debt level of the typical American has risen far out of proportion to his or
her income.”).
16. On the development and implications of risk-based pricing—see Mark Furletti (2003)
“Credit Card Pricing Developments and Their Disclosure” (January 2003) (Fed. Res. Bank of
Philadelphia, Payment Cards Center, Discussion Paper, available at <http://www.phil.frb
.org/pcc/papers/2003/CreditCardPricing_012003.pdf>. For the eventual reduction in inter-
est rates—see below, II.B.2.
17. On cards vs. cash—see Sullivan, Warren, and Westbrook, above note 1, at 108; Evans and
Schmalensee, above note 5, at 25–6. For the 2008 figures—see 2011 Statistical Abstract, above
note 1 (tbl. 689 lists the number of households: 117,181,000, and median household income:
$50,303; tbl. 1187 lists the total number of credit cards and the total credit card purchase
volume).
c re di t card s 61
cards, 73 percent of households had at least one credit card in 2007. The
average monthly household charge was over $1,500 in 2008, as compared
to only $165 (in current dollars) in 1970. And the ratio of charges-to-income
grew from just under 4 percent in 1970 to about 35 percent in 2008. Credit
cards’ piece of the total consumer spending pie has increased substantially.
A growing body of evidence suggests that they have also contributed to the
increase in the size of the consumer spending pie.18
Not only do credit cards encourage spending, they encourage borrowing
as well. As noted by Sullivan, Warren, and Westbrook:
Credit card debt has become as much a part of American life as has the credit
card itself. . . . Of the three-quarters of all households that have at least one
credit card, three out of four of them also carry credit card debt from month
to month. . . . Increasingly, . . . [Americans] do not pay [with their credit cards]—
they finance. Quietly, without much fanfare, Americans have taken to buying
school shoes and pizza with debt—and paying for those items over months or
even years.19
18. On the growth of the credit card industry—see generally Evans and Schmalensee, above note
5, at 235–6 (“Between 1971 and 1997, the number of cards in circulation increased by more
than 900 percent, while the number of households increased by only 54 percent. The total
dollar value of credit card transactions increased by 2,630 percent in that same period while
personal consumption expenditures increased by 125 percent. Finally, outstanding balances
increased by 2,700 percent while total consumer credit outstanding increased by 140 per-
cent. . . . Payment cards have grown at the expense of other means of payment and other
sources of credit.”) For the 1970 figures—see Evans and Schmalensee, above note 5, at 85–7,
240 (for 1970, Evans and Schmalensee report an average household charge of $125 in 1998
dollars, which is equivalent to $165 using the CPI). For the 2007 figure—see Bucks et al.,
above note 5, at A1–A55 (based on the 2007 Survey of Consumer Finances). For the 2008
figures—see 2011 Statistical Abstract, above note 1, tbls. 689, 1187 (for 2008 figures) (the aver-
age monthly household charge was calculated by dividing annual credit card purchase
volume—$2,153 bil., from tbl. 1157—by the number of households—117,181,000, from tbl.
689—and then by 12, to get monthly figures; the charges to income ratio was calculated by
dividing the average household charge by the median household income—$50,303, from tbl.
689). On credit cards and consumer spending—see Lloyd Klein, It’s In The Cards: Consumer
Credit and the American Experience (Praeger, 1999) (“Credit cards facilitated the rise of consumer
spending for consumer products or services.”); Elizabeth C. Hirschman, “Differences in Con-
sumer Purchase Behavior by Credit Card Payment System”, J. Consumer Res., 6 (1979),58
(people who own more credit cards make larger purchases per department store visit); Richard
A. Feinberg, “Credit Cards as Spending Facilitating Stimuli: A Conditioning Interpretation”,
J. Consumer Res., 12 (1986), 384 (restaurant tips are larger when payment is by credit card);
Drazen Prelec and Duncan Simester, “Always Leave Home Without It: A Further Investiga-
tion of the Credit-Card Effect on Willingness to Pay”, Marketing Letters, 12 (2001), 5 (respond-
ents offered significantly higher prices for Celtics and Red Sox tickets when paying by credit
card). These studies preclude a liquidity constraints explanation for the credit card effect. See,
e.g., id. at 10.
19. Sullivan,Warren, and Westbrook, above note 1, at 110–11. See also David B. Gross and Nicho-
las S. Souleles, “Do Liquidity Constraints and Interest Rates Matter for Consumer Behavior?
Credit cards are now the leading source of unsecured consumer credit/debt.
And consumers aren’t the only ones who use credit card financing; many
self-employed owners of small businesses turn to high-interest credit card
debt to finance their businesses.20
Total credit card borrowing in 2008 amounted to $976 billion. The aver-
age credit card debt per household in the U.S. was over $8,300 in 2008. If
we focus on the 73 percent of households with credit cards, average debt per
household rises to over $11,400. The average debt among the 60 percent of
households that carry credit card debt was over $19, 000 in 2008.The median
credit card debt-to-income ratio was 16.5 percent across all households,
22.7 percent for households with credit cards, and 37.8 percent for house-
holds that carry credit card debt. Clearly, credit card debt imposes a substan-
tial burden on many households.21
It is not surprising, then, that credit card debt plays a notoriously import-
ant role in consumer bankruptcy. Credit card defaults are highly correlated
with personal bankruptcies. Based on their empirical investigation of con-
sumer bankruptcy filings, Sullivan,Warren, and Westbrook conclude that “[a]
s the fastest growing proportion of consumer debt, credit card debt has led
the way to bankruptcy for an increasing number of Americans. . . . ” Three
independent statistical studies, by Ronald Mann, by Atif Mian and Amir
Evidence from Credit Card Data”, Q. J. Econ., 117 (2002), 149, 151 (Of all households with at
least one bankcard “at least 56 percent—a remarkably large fraction—are borrowing on their
bankcards, that is, paying interest, not just transacting.”These figures, which are based on SCF
data, significantly understate the percentage of households with credit card debt, since SCF
households substantially underreport their credit card debt.); Klein, above note 18, at 29 (“[t]
he ‘me generation,’ actualized through credit card utilization, was transformed into a debt
carrying ‘greed generation’ wanting and buying everything in sight.”)
20. See Sullivan, Warren, and Westbrook, above note 1, at 115–17; Evans and Schmalensee, above
note 5, at 34, 103–7.
21. For total and average per-household credit card debt—see 2011 Statistical Abstract, above
note 1, tbl. 689, 1187 (average per-household debt was calculated by dividing total credit
card borrowing by the number of households—117,181,000, from tbl. 689). For the per-
centage of households with credit cards and for the share of those households that carry
debt—see Bucks et al., above note 5, at A1–A55 (based on the 2007 Survey of Consumer
Finances, 73 percent of families had at least one credit card, and among those families 60.3
percent carried a balance). The debt-to-income ratios were calculated by dividing the aver-
age debt figures above by the median household income—$50,303. See 2011 Statistical
Abstract, above note 1, tbl. 689. The debt-to-income ratio has been growing. See Sullivan,
Warren, and Westbrook, above note 1, at 18 (“[R]eal consumer debt has risen dramatically
over a long period during which real incomes for many people have stayed the same or
declined.”) See generally Durkin, above note 1, at 202 (“Much of the growth of consumer
credit in recent years has been in the form of revolving credit, of which credit card credit
is the largest component.”); Sullivan, Warren, and Westbrook, above note 1, at 129 (“As the
c re di t card s 63
Sufi, and by Michelle White, suggest a causal relationship between credit

card debt and consumer bankruptcy filings.22
D. Market Structure
1. Participants
The credit card market is divided among the major credit card brands: Visa,
MasterCard, American Express, and Discover. The bankcard brands, Visa and
MasterCard, share a common and more complex structure. Until recently, they
were joint ventures of banks, comprising thousands of distinct issuers. In 2006,
MasterCard registered as a private-share corporation, owned by its member
banks.Visa followed suit in 2008.While many of these bank issuers operate only
at the local level, a significant number of others participate at the regional and
national levels. In the 1990s, a new group of players entered the credit card
scene; non-bank issuers, such as AT&T. While technically these non-bank issu-
ers are necessarily affiliated with a Visa or MasterCard issuing bank and the
issued credit card is actually a co-brand card, as in the case of AT&T and Mas-
terCard, the major strategic decisions are undertaken by the non-bank issuer.23
Regarding the bankcard brands, it is interesting to note that the Visa and
MasterCard associations are quite decentralized. Lawrence Ausubel observed
fastest growing proportion of consumer debt, credit card debt has led the way to bank-
ruptcy for an increasing number of Americans. . . . ”).
22. For the bankruptcy filings study—see Sullivan,Warren, and Westbrook, above note 1, at 129. See
also Sullivan, Warren, and Westbrook, above note 1, at 119–20; Evans and Schmalensee, above
note 5, at 5. For the three statistical studies—see Ronald J. Mann, Charging Ahead: The Growth
and Regulation of Payment Card Markets around the World (Cambridge University Press, 2006); Atif
R. Mian and Amir Sufi, “Household Leverage and the Recession of 2007–09”, IMF , 58 (2010),
74–117 (“Overall, the statistical model shows that household leverage growth and dependence
on credit card borrowing as of 2006 explain a large fraction of the overall consumer default,
house price, unemployment, residential investment, and durable consumption patterns during
the recession.”); Michelle J.White,“Bankruptcy Reform and Credit Cards” (2007) J. Econ. Persp.,
21 no. 4 (2007) (“From 1980 to 2004, the number of personal bankruptcy filings in the United
States increased more than five-fold, from 288,000 to 1.5 million per year. . . . I argue that the
main explanation is the rapid growth in credit card debt, which rose from 3.2 percent of U.S.
median family income in 1980 to 12.5 percent in 2004.”). But see Board of Governors of the
Federal Reserve System, Report to the Congress on Practices of the Consumer Credit Industry
in Soliciting and Extending Credit and their Effects on Consumer Debt and Insolvency, June
2006, 3, 26 (“Though the percentage of families holding credit cards has steadily risen, the
household debt service burden has only modestly increased and the majority of households pay
their revolving debt on time.”; and noting that most bankruptcies in the United States are actu-
ally triggered by life crises such as divorce, job loss, and uninsured illness.); Todd J. Zywicki,
“The Economics of Credit Cards”, Chap. L. Rev., 3 (2003), 79, 82 (doubting the link between
credit cards and bankruptcy rates).
23. See Evans and Schmalensee, above note 5, at 4, 48–9, 75–7. See also Ausubel, above note 11,
at 51 (there are more than 4,000 card-issuing banks).
that “most relevant business decisions are made at the level of the issuing
bank [rather than at the Visa or MasterCard organizations level]. Individual
banks own their cardholders’ accounts and determine the interest rate,
annual fee, grace period, credit limit, and other terms of the account.”24
Even so, the Visa and MasterCard organizations play an important role.
They set the interchange fee for the transfer from the merchant’s bank to
the card-issuing bank. They promote the company’s brand name through
advertising. And they lead the competition with the other bankcard brand
as well as with the non-bankcard brands.25
2. Competition
There are two intertwined levels of competition within the credit card
industry; brand-level competition and issuing-level competition.
At the upper level, the major credit card brands—especially Visa, Master-
Card, American Express, and Discover—compete among themselves.Visa is
the industry leader in terms of both charge volume and credit extended,
with MasterCard following closely behind. American Express and Discover
occupy the more distant third and fourth places, respectively.26 The evidence
regarding the intensity of competition at this level is mixed. While the four
brands clearly compete against each other, the series of antitrust challenges
against Visa and MasterCard suggests that the leading brands have taken
steps to limit competition at the network level.27
Competition at the issuing level is more robust, although there is evidence
that competition even at this level is less than perfect. While it may appear
that thousands of banks, along with American Express and Discover (as issu-
ers), compete for consumers, the fact is that a relatively small number of large
banks control most of the market. There is also mixed evidence about the
competitive effect of switching costs. While the economic transaction costs
24. Ausubel, above note 11, at 51.

25. Id. See also Evans and Schmalensee, above note 5, at 199.
26. See Nilson Report, February 2010. In 2009, the U.S. credit card market was divided among
these issuers as follows: Visa—43.4 percent with $764.2 billion in purchase volume;
MasterCard—27.1 percent with $476.9 billion in purchase volume; American Express—23.8
percent with $419.8 billion in purchase volume; and Discover—5.7 percent with $100.4 bil-
lion in purchase volume. Focusing on credit extended by the four major brands, the 2009
market shares were: Visa—47.4 percent with $366.05 billion of outstanding credit; Master-
Card—34.7 percent with $267.57 billion of outstanding credit; American Express—11.1 per-
cent with $86.06 billion of outstanding credit; and Discover—6.8 percent with $52.51 billion
of outstanding credit.
27. For a summary of the different antitrust challenges faced by Visa and MasterCard over the
years—see Evans and Schmalensee, above note 5, ch. 11.
c re di t card s 65
of obtaining a new card are fairly insignificant for many borrowers, which
would lead one to think that consumers would switch cards frequently, there
is evidence that psychological switching costs and simple inertia restrict
switching rates. One study estimated total switching costs at $150. Moreover,
issuers purposefully increase switching costs with loyalty programs.28
Competition is also affected by the availability of information and by bor-
rowers’ ability to process the information needed to comparison-shop. At first
glance, it appears that there’s quite a bit of information available, especially through
the internet. But it’s not clear how many consumers are actually able to find and
process this information. As will be explained in the pages that follow, the credit
card product is multidimensional and complex. Comparison shopping between
several multidimensional, complex products is a daunting task for many consum-
ers. Finally, the profitability of credit card issuers consistently exceeds the average
profitability in the banking industry, leading some commentators to conclude
that competition in the credit card market is imperfect. And, allegations of coor-
dinated action have been made against the major issuers.29
In what follows, I largely sidestep the unresolved question about the level
of competition among credit card issuers. I show that the central failure in
the credit card market—the behavioral market failure—leads to inefficien-
cies that cannot be cured by even perfect competition. This justifies placing
the credit card industry under scrutiny, even if it is subject to intense com-
petition that dissipates any supra-competitive rents.
II. The Credit Card Contract

Credit card pricing patterns are indicative of a behavioral market failure.
In this section, we’ll take a look at the relevant features of credit card
28. The Second Circuit noted that “competition . . . is robust at the issuing level.” See United States v.
Visa U.S.A., Inc., 344 F.3d 229, 240 (2nd Cir. 2003).Yet the nine largest issuers control approxi-
mately 90 percent of the market. See CFPB, CARD Act Factsheet, above note 4. On switching
costs—see Evans and Schmalensee, above note 5, at 234–5 (low economic switching costs); Haiyan
Shui and Lawrence M. Ausubel “Time Inconsistency in the Credit Card Market” (2004) Working
Paper, <http://ssrn.com/abstract=586622> (high psychological switching costs).
29. On profitability—see Federal Reserve Board, Report to the Congress on the Profitability of
Credit Card Operations of Depository Institutions, June 2009 (“Although profitability for the
large credit card banks has risen and fallen over the years, credit card earnings have been
consistently higher than returns on all commercial bank activities.”). On allegations of
coordination—see In re Currency Conversion Fee Antitrust Litigation (MDL No. 1409) (class set-
tlement approved); Ross et al. v. Bank of America, N.A. (USA), No. 05-cv-7116, MDL No. 1409
(S.D.N.Y.) (settlement reached with a subgroup of the defendants).
pricing.30 Then we’ll explore competing rational-choice and behavioral-

economics explanations for these pricing schemes.
A. Complexity
The common credit card contract is complex and multidimensional, with
numerous fees and interest rates. Credit card fees can be divided into service
fees and penalty fees. The service fees include application fees, set-up fees,
annual fees, membership fees, participation fees, cash-advance fees, balance-
transfer fees, foreign-currency-conversion fees, credit-limit-increase fees,
expedited-payment or phone-payment fees, no-activity fees, fees for stop
payment requests, fees for statement copies, fees for replacement cards, and
wire-transfer fees. And then there are penalty fees, including late fees, over-
limit fees, and returned-check (NSF—No Sufficient Funds) fees. According
to one industry source, “an average of 9 cardholder fees and costs [are]
found in cardholder fee disclosures.”31
Now let’s look at interest rates. In addition to the main, long-term interest
rate for purchases, there are introductory (teaser) rates, rates on balances trans-
ferred from other cards, rates on cash advances, and default interest rates.32
Most of these fees and interest rates are themselves variable and multi-
dimensional. For example, interest rates are often variable rates that change
over time, tracking the movements of a certain index. Introductory rates last
for specified time periods, as do default interest rates. Late fees depend
(sometimes) on the magnitude of the balance. Etc.33
30. I focus on pricing on the consumer side of the credit card market. It should be noted, how-
ever, that the credit card market—being a two-sided network market—exhibits interesting
pricing patterns also on the merchant side of the market and between the two sides of the
market (namely, the interchange fee that acquirers pay to issuers). For a survey of the literature
that studies these other aspects of credit card pricing—see Sujit Chakravorti, “Theory of
Credit Card Networks: A Survey of the Literature” Rev. Network Econ. 2 (2003), 50.
31. CardFlash, “Penalty Fees,” January 20, 2009. See also Furletti, above note 16, at 5, 10–13; Fed-
eral Reserve Board, “Credit Cards—Fees”, <http://www.federalreserve.gov/creditcard/fees
.html> (last visited June 24, 2011); Elizabeth Renuart and Diane E. Thompson, “The Truth,
The Whole Truth, and Nothing but the Truth: Fulfilling the Promise of Truth in Lending”,
Yale J. Reg., 25 (2008), 181, 192–4.
32. See Furletti, above note 16, at 14 (“the number of APRs that can be applied to the balances on
an account has increased dramatically over the past 10 years (e.g., purchase APR, promotional
APR, cash APR, balance transfer APR))”; Federal Reserve Board, “Credit Cards—Interest
Rates”, <http://www.federalreserve.gov/creditcard/rates.html> (last visited June 24, 2011).
33. See Federal Reserve Board, “Credit Cards—Fees”, <http://www.federalreserve.gov/credit-
card/fees.html> (last visited June 24, 2011); Federal Reserve Board, “Credit Cards—Interest
Rates”, <http://www.federalreserve.gov/creditcard/rates.html> (last visited June 24, 2011).
c re di t card s 67
In addition, the amount a consumer pays in interest depends on the bal-

ance that she carries, and these balances are calculated using complex for-
mulas. In the late 1990s, issuers moved from monthly to daily compounding
of interest, adding finance charges to the balance each day. This change had
the effect of increasing the effective APR by as much as 10 to 20 basis points.
Moreover, before the practice was banned by the CARD Act of 2009, issu-
ers used another complex calculation called double-cycle billing. Double-
cycle billing effectively eliminated the grace period (the interest-free period
that consumers who pay their bill in full receive from the time they make a
purchase until the date their payment is due) for consumers who made
partial payments the month after making a full payment or who made a
partial payment the month following a zero-balance.34
When a cardholder has multiple balances with the same issuer (for exam-
ple, one for transferred balances, one for cash advanced, and one for regular
purchases), the credit card contract specifies how payments will be allocated
among the different balances. This payment-allocation method affects total
interest payments. Before the practice was banned by the CARD Act of
2009, many issuers allocated payments to low-interest balances first, thus
maximizing the finance charges.35 The minimum payment requirement also
affects total interest payments, since slower repayment results in more inter-
est paid overall. A cardholder who wants to predict his or her total interest
payments faces a daunting task.
With respect to contingent penalty fees and rates, the contingency that
triggers the fee or rate needs to be specified. For example, when is a pay-
ment considered late—when it arrives on the due date or when it arrives
34. See Furletti, above note 16, at 15–16 (describing the balance calculation methods); Federal
Reserve Board, “What You Need to Know: New Credit Card Rules Effective February 22”,
<http://www.federalreserve.gov/consumerinfo/wyntk_creditcardrules.htm> (last visited
June 24, 2011) (hereinafter FRB, “Credit Card Rules, February 22”) (describing the new
CARD Act rules).
35. Furletti, above note 16, at 14–15 (describing the payment allocation methods used before
they were banned by the CARD Act); PEW Report, Safe Credit Card Standards 1 (2009),
available at <http://www.pewtrusts.org/our_work_detail.aspx?id=616>. (A study cover-
ing 400 cards offered online by the top 12 issuers, which control more than 88 percent of
outstanding credit card debt, found that 100 percent of cards in the study allowed the issuer
to apply payments in a manner which, according to the Federal Reserve, is likely to cause
substantial monetary injury to consumers.) Furletti concludes that these balance calcula-
tion and payment allocation methods have a material effect on issuer revenues, increasing
effective yields without affecting the disclosed nominal APRs (Furletti, above note 16, at
14–16). FRB, “Credit Card Rules, February 22,” above note 34 (describing the new CARD
Act rules).
before a certain hour on the due date? What if the due date falls on a week-
end or holiday?
When is a default interest rate triggered? In the not so distant past, many
credit card contracts included “universal default” clauses, triggering penalty
rates based on a host of factors, such as a change in the cardholder’s credit score
or a failure to pay a utility bill on time. The CARD Act, while not explicitly
banning “universal default,” restricts issuers’ ability to increase rates.36
Many cards also provide a long list of ancillary benefits, including loy-
alty rewards programs (frequent-flyer miles, cash-back, and the like), rental
car insurance, and more. These programs and benefits are themselves quite
complex. For example, the program rules detail how miles or points are
accumulated, how they can be redeemed, and when they expire if not
used. Finally, protection against fraud is of paramount importance to many
cardholders. Different issuers offer different mechanisms for protecting
their cardholders—another complex benefit.
All these complex details affect the costs and benefits of the credit card
product.
B. Deferred Costs
The complex and multidimensional credit card contract provides bene-
fits to consumers while imposing costs on them. These benefits and
costs are not randomly distributed across the many dimensions of the
contract. Instead, benefits are provided through short-term, more sali-
ent dimensions while costs are imposed through long-term, less salient
dimensions. With pricing terms, for example, long-term contingent
price elements are over-priced while short-term, non-contingent price
elements are under-priced. In other words, benefits are accelerated,
costs are deferred.
1. Short-Term Benefits
a. No Annual or Per-Transaction Fees
For an issuer, the most straightforward way to cover the fixed costs associ-
ated with establishing and maintaining a credit card account would be to
36. See Furletti, above note 16, at 8–9 (quoting the language of a common “universal default
clause” from a credit card contract); FRB, “Credit Card Rules, February 22,” above note 34
(describing the new CARD Act rules).
c re di t card s 69
charge an annual fee—which is, in fact, what they used to do. Nowadays,
however, it is common for issuers to charge no annual fee.37
Similarly, issuers incur the costs of handling the transactions that con-
sumers charge to their credit cards. Yet most credit card pricing does not
include any per-transaction fee. In fact, when considering the benefits or
rewards programs associated with many credit cards, issuers are setting
negative per-transaction fees. The proliferation of membership-rewards pro-
grams, frequent-flyer miles, car rental and luggage insurance, discounts on
future purchases, and cash-back grants all point to the fact that competitive
forces are at play on this dimension of the credit card contract.
Even though they incur positive costs in maintaining credit card accounts
and processing transactions, issuers commonly set a zero—or even a
negative—price for these services. These observations suggest that issuers
are charging below-margin-cost prices on the annual fee and per-transac-
tion fee dimensions of the credit card contract.
But the story is more complex. The preceding account, like the rest of
this chapter, focuses on the issuer-consumer contract. This contract sets a
low, even negative, price on the per-transaction dimension. A broader analy-
sis would consider also revenue that issuers obtain from merchants, via the
interchange fee. (The interchange fee is the percentage—usually 2
percent—that is transferred from the merchant to the issuer on each credit
card purchase.) The point is that issuers choose not to charge consumers for
the transacting service that the credit card provides. Moreover, while mer-
chants may raise prices to compensate for the “tax” that they pay to issuers,
cardholders share the burden of these higher prices with customers using
other payment methods.38
b. Teaser Rates
Teaser rates are the low, introductory rates that many credit card issuers
offer, typically for a period of six months. After that, a higher long-
term interest rate kicks in. Some cards even offer a zero-interest rate
37. See Furletti, above note 16, at 9–10. (In the mid-1990s issuers eliminated annual fees “in prime
portfolios not associated with a rewards program.”) See also Evans and Schmalensee, above
note 5, at 159.
38. This would not have been true if merchants charged higher prices for credit card purchases.
In fact, such a price differential would constitute a de facto per-transaction fee for credit card
transactions. Charging different prices as a function of the payment method is, however, quite
rare (and, until recently, either illegal or effectively prevented by card networks’ no-discrimin-
ation rules). See, e.g., Chakravorti, above note 30, at 55–6.
during the introductory period. A zero-interest rate on balance transfers

is also common.39
Enjoying a low (or no) interest rate is clearly a benefit to consumers, even
if this low rate expires after a certain period. Teaser rates also hold the pros-
pect of even greater benefit if consumers transfer their balance to a new
card offering a low teaser rate as soon as the introductory period on the
current card expires and the post-introductory rate kicks in. The result
would be free credit for an indefinite period.While balance transfers—from
a card with an expiring teaser rate to a new card with a new teaser rate—do
occur, available evidence suggests that they are not as prevalent as one might
expect. Rather, substantial borrowing is done at the post-introductory rates.
In fact, most borrowing is done at the high post-promotion rates, rather than
at the low teaser rates.40
Teaser rates highlight the disparity between credit card interest rates and
the underlying cost of funds. Clearly, the cost of funds is greater than zero,
yet zero-percent teaser rates are not uncommon. Teaser rates also imply a
sharp discontinuous increase at the end of the introductory period. It is dif-
ficult to argue that such a rate increase corresponds to changes in the issuer’s
cost of funds.
2. Long-Term Costs
a. Long-Term Interest Rates
A central element of credit card pricing is the interest rate charged on credit
card debt—the long-term interest rate on purchases that kicks in after the
short-term or introductory/teaser interest rate period has expired. (It’s
important to note that these interest rates on purchases can be different
from the long-term interest rates on transferred balances or cash advances.)
Credit card contracts set high interest rates.The average credit card inter-
est rate was 13.30 percent in 2007 and 12.08 percent in 2008.41 These high
39. See CardFlash, “Dec Card Offers,” December 23, 2008. (“Zero percent teaser rates for 12
month periods returned in force in Dec. 2008.”) See also Stefano DellaVigna and Ulrike
Malmendier, “Contract Design and Self-Control: Theory and Evidence”, Q. J. Econ., 119
(2004), 353, 377–8; Ausubel, above note 10, at 262.
40. See Gross and Souleles, above note 19, at 171, 179. See also Ausubel, above note 10, at 263 (“a
substantial portion of credit card borrowing still occurs at postintroductory interest rates”;
“finance charges paid to credit card issuers have not dropped as much as the introductory
offers might suggest”); David I. Laibson et al., “A Debt Puzzle,” in Philippe Aghion et al.
(eds.), Knowledge, Information, and Expectations in Modern Macroeconomics: In Honor of Edmund, S.
Phelps 228–9 (Princeton University Press, 2003) (finding that consumers pay high effective
interest rates “[d]espite the rise of teaser interest rates”).
41. See Federal Reserve Statistical Release, “Consumer Credit” (Sep. 8, 2010).
c re di t card s 71
interest rates are not surprising. It is natural for unsecured debt, like credit
card debt, to carry higher interest rates than secured debt, such as mortgages,
home equity lines of credit, and car loans.
In the 1980s and early 1990s, real concern surfaced about the magnitude
of credit card interest rates. This concern was based, in large part, on the
“stickiness” of these rates—the evidence that credit card interest rates did
not track changes in the underlying cost of funds. David Evans and Richard
Schmalensee, both Visa consultants, note: “Not only are credit card interest
rates high, they do not always move as quickly as other interest rates in
response to changes in the cost of the funds that banks raise to support their
lending activities.”42
The concern about credit card interest rates has dissipated. Since the late
1990s, competition had begun to focus on the interest rate dimension, driv-
ing interest rates down and forcing a stronger correlation with the under-
lying cost of funds.43
But while the interest rates themselves are becoming more competitive,
there is still concern about interest payments. As explained above, interest
payments are not only a function of interest rates; they also depend on how
balances are calculated, how payments are allocated to different balances,
and on the amount of the minimum payment.
Until the CARD Act banned this practice, most issuers calculated bal-
ances using the double-cycle billing method. While the main concern with
this method is its complexity and opacity, it also generates elevated interest
payments in certain circumstances. Similarly, until banned by the CARD
Act, issuers commonly allocated cardholders’ payments to the balance bear-
ing the lowest interest rate first. The result was increased overall interest
payments.44
42. Evans and Schmalensee, above note 5, at 248. See also Ausubel, above note 10, at 261
(“throughout the remainder of the 1980s, credit card interest rates displayed a profound unre-
sponsiveness to changes in the cost of funds”).
43. See Furletti, above note 16, at 2, 3, 29 (“from 1992 to 2001, however, the average interest rate
that issuers charged revolving customers fell 320 basis points, from 17.4 percent to 14.2 per-
cent. Issuer markup, a metric that normalizes for funding costs by subtracting the six-month
Treasury bill rate from the average APR, decreased 330 basis points during the same period”;
“an overall decrease in the average APR, coupled with an increase in the number of
lower income credit users, suggests that the average rate decrease for many cardholders was
even more pronounced than the average APR indicates”); Figure 3: Average Credit Card
Markup (Average Credit Card APR minus 6m Treasury Bill Rate)—declining since 1992).
44. See Section A above.
The minimum required payment also has a large effect on the overall
interest paid by cardholders. Credit card issuers often require only a very
small minimum monthly payment, even for large outstanding balances. As
long as cardholders do not default, lower monthly payments “increase total
revenues by increasing the time it takes to repay the loans and hence the
total interest eventually repaid.”45
b. Penalty Fees and Default Interest Rates
Credit card issuers typically collect sizeable fees from consumers who either
run late on their monthly payments or exceed the credit limit. Importantly,
the magnitude of these penalties is often measured in fixed-dollar amounts—
typically around $35—regardless of the degree of deviation from the credit
line or tardiness in making the payment.46 Thus, for example, a cardholder
might pay a $35 penalty if she misses the due date on a $10 balance by a few
days. Hard-pressed to justify such a fee structure, some issuers have begun to
offer gradation of late fees, such that smaller fees are imposed for tardiness
in paying off a smaller balance.
Paying late or exceeding the credit limit is not a fringe phenomenon.
According to a Government Accountability Office study of the six largest
issuers 35 percent of accounts were assessed late fees and 13 percent of accounts
were assessed over-limit fees. Late fees and over-limit fees are a major source
of revenue for credit card issuers, reaching almost $20 billion or approximately
10 percent of total card revenues in 2008. Issuers have also been shortening
grace periods to further enhance revenues from penalty fees.47
45. Sullivan,Warren, and Westbrook, above note 1, at 247–8. See also FRB, “Credit Card Lending:
Account Management and Loss Allowance Guidance,” SR Letter 03–1, 3–4, January 8, 2003,
available at <http://www.federalreserve.gov/boarddocs/srletters/2003/sr0301.htm>.
(“Competitive pressures and a desire to preserve outstanding balances have led to a general
easing of minimum payment requirements in recent years” and “In many cases, reduced mini-
mum payment requirements in combination with continued charging of fees and finance
charges have extended repayment periods well beyond reasonable time frames.”); Tamara
Draut and Javier Silva, Borrowing to Make Ends Meet:The Growth of Credit Card Debt in the ’90s,
p. 37 (Dēmos, 2003) (hereinafter “Dēmos”).
46. See CardFlash, “Fee Factor 08,” January 14, 2009. (Between 1994 and 2008, average late pay-
ment fees increased from $12.52 to $35.36, and average over-limit fees increased from $12.74
to $35.91 (<http://www.carddata.com>)). See also Pfennig v. Household Credit Services, Inc.,
No. 00–4213 (6th Cir., July 2, 2002) ($29 over-limit charge for every month the balance
remained over the credit limit, regardless of the degree of deviation from the credit limit);
Sullivan, Warren, and Westbrook, above note 1, at 23 (“penalty fees added on at $50 a pop”).
47. On the number of cardholders paying late fees—see U.S. Government Accountability Office,
Credit Cards: Increased Complexity in Rates and Fees Heightens Need for More Effective
Disclosures to Consumers, 5 (2006) (hereinafter “GAO Complexity Report”). See also Fur-
letti, above note 16, at 11 (Cardweb estimated that half of cardholders had made a late
c re di t card s 73
It’s hard to justify, based on the extra costs of extending the loan period
or raising the loan limit, the high fees that issuers charge for late payments
and for exceeding the credit limit—even accounting for the potentially
heightened risk of accommodating a consumer who fails to pay on time or
to remain within the specified credit line. This disparity between price and
cost is especially striking when the late and over-limit fees are set at fixed-
dollar amounts, regardless of the tardiness of the payment or the magnitude
of the deviation from the credit limit.48
It should be noted that the CARD Act now requires that penalty fees be
“reasonable and proportional” to the violation of the account terms in
question. These restrictions have led to a substantial reduction in the mag-
nitude of assessed fees.The CARD Act also prohibits issuers from automati-
cally enrolling cardholders in over-the-limit programs, where cardholders
can exceed their credit limit but are then charged an over-limit fee. Issuers
must now obtain express consent from cardholders who wish to enroll in
such programs. This switch, from an opt-out to an opt-in regime, has led
many issuers to drop the over-limit fee.49
payment in 2001). On revenues from late fees—see CardFlash, “Fee Income,” January 11, 2010
(total penalty fee income in 2008 amounted to $19 billion; while late fees and over-limit fees
account for the lion’s share of the $19 billion figure, the $19 billion includes also returned-
check (NSF) fees, and currency conversion fees); CardFlash, “Fee Factor 08,” January 14, 2009
(total revenues in 2008 were $169 billion; the percentage of total revenues figure was calcu-
lated by dividing total penalty fee income by total revenues). See also “National Consumer
Law Center, Truth in Lending” 27 (2002 Cumulative Supplement). (“Over-limit fees are a
major source of revenue for many credit card issuers.”) Penalty fees have been growing rapidly
since 1996 when the Supreme Court extended the Marquette rule to include late and over-
limit fees. See Smiley v. Citibank, 517 U.S. 735 (1996). See also CardFlash,“Fee Factor 08,” Janu-
ary 14, 2009 (“The rise in fees for a late payment, being over-limit, making a cash advance/
balance transfer or an expedited payment were triggered in the wake of the 1996 Smiley deci-
sion by the U.S. Supreme Court which exempted national card issuers from state restrictions
on card fees.”); Dēmos, above note 45, at 35 (late fees are the fastest growing source of reve-
nues for issuers). On the shortening of grace periods—see Office of Comptroller of Currency,
Advisory Letter: Credit Card Practices, 3, September 14, 2004, available at <http://www.occ
.treas.gov/Advlst04.htm> (describing issuer practices: “Credit card issuers may take other
actions that also effectively increase the cost of credit for some consumers, such as shortening
the due date for receipt of payment.”).
48. See Nadia Ziad Massoud, Anthony Saunders, and Barry Scholnick, “The Cost of Being Late:
The Case of Credit Card Penalty Fees” (2006) AFA 2007 Chicago Meetings Paper available at
<http://ssrn.com/abstract=890826> (penalty fees are correlated with risk factors, but also
with market share—consistent with the rent extraction theory).
49. See FRB, “Credit Card Rules, February 22,” above note 34 (describing the new CARD Act
rules); CFPB, CARD Act Fact Sheet, above note 4 (describing the new CARD Act rules and
their effects, specifically the reduction in late fees and the “virtual disappearance” of over-
limit fees).
Like penalty fees, default interest rates are also a major source of revenue
for issuers. These default rates, which exceed 20 percent and sometimes
even 30 percent, are easily triggered by such contingencies as a late payment
or a charge that exceeds the credit limit. The rates were even more easily
triggered before universal default was substantially curtailed by the CARD
Act. The Center for Responsible Lending estimated that almost 11 percent
of all outstanding credit card balances carried penalty pricing in 2008. And
a 2008 study by the PEW Charitable Trusts’ Safe Credit Cards Project found
that imposing the typical penalty rate costs nearly $500 for the typical,
re-priced account. Aggregated across U.S. cardholders, this amounts to over
$7 billion a year.50
c. Other Fees
The last category of long-term costs comprises a number of non-penalty
fees, such as the currency conversion, cash advance, balance transfer, no
activity, and convenience or telephone payment fees. These fees bring in
substantial revenue for issuers: approximately $10 billion annually. Two of
them—the no-activity fee and the convenience fee—have been recently
banned by the CARD Act. Another—the currency-conversion fee—was
the subject of a $336 million class action settlement. Other terms in the
credit card contract, while not direct fees, impose long-term, contingent
costs on cardholders. Mandatory arbitration clauses, the subject of recent
legal and public scrutiny, are a prime example.51
50. On the magnitude of default interest rates—see PEW Report, above note 35, at 1 (in a
recent study covering 400 cards offered online by the top 12 issuers, which control more
than 88 percent of outstanding credit card debt, the median allowable penalty interest
rate was 27.99 percent per year); William Weeks, “An Analysis and Critique of Retroac-
tive Penalty Interest in the Credit Card Market” (2007) mimeo (documenting penalty
rates of 20 percent to 30 percent and above). For the number of accounts carrying pen-
alty pricing—see Center for Responsible Lending, “Priceless or Just Expensive? The Use
of Penalty Rates in the Credit Card Industry,” (2008). For the cost of penalty rates to
consumers—see PEW Charitable Trusts, “Safe Credit Cards Project: Curing Credit Card
Penalties” (2009) 1, available at <http://www.pewtrusts.org/our_work_detail .
aspx?id=616>.
51. On fee revenues—see CardFlash, “Fee Income,” January 11, 2010 (income from cash
advance fees alone amounted to $8 billion in 2009); Furletti, above note 16, at 12 (Con-
venience and service fees “are generally priced to provide attractive profit margins.”). The
so-called fee-harvester cards should also be mentioned. See Report, Fee-Harvesters: Low-
Credit, High-Cost Cards Bleed Consumers (National Consumer Law Center, 2007). For new
CARD Act restrictions—see FRB, “What You Need to Know: New Credit Card Rules
Effective August 22”, <http://www.federalreserve.gov/consumerinfo/wyntk_
creditcardrules.htm> (last visited June 24, 2011) (describing the new CARD Act rules,
including a ban on inactivity fees); Credit Card Accountability, Responsibility and Disclosure
Act of 2009, Pub. Law 111–24, Sec. 102 (May 22, 2009) (restricting convenience fees). On the
c re di t card s 75
III. Rational-Choice Theories and Their Limits

Why does the common credit card contract look the way it does? What
explains its complexity and deferred-cost features? We’ll explore these ques-
tions by first examining rational-choice, efficiency-based accounts. As we’ll
see, rational-choice theories explain certain contract design features in cer-
tain circumstances, but leave an explanatory gap.We’ll fill that gap in Part IV
with a behavioral-economics theory.
A. Complexity
1. Interest Rates
The common credit card contract specifies different interest rates, from the
basic interest rate for purchases to an interest rate for balance transfers to
an interest rate for cash advances. In a rational-choice model, separating the
different balances and matching different interest rates to each balance
would be efficient if the different balances were associated with different
risk levels. Forcing a single, common interest rate would prevent efficient
risk-based pricing and lead to cross-subsidization between different groups
of cardholders.
Balance calculation and payment-allocation methods also affect interest
payments and add to the complexity of the credit card contract. Identifying
a rational-choice, efficiency-based explanation for the complex-balance
calculation and payment-allocation methods is a difficult task. It is not clear
how double-cycle billing enhances efficiency. Similarly, while allocating
payments to low-interest balances first increases issuers’ revenues, it is not
clear how it increases overall efficiency.
currency conversion fee class action—see In re Currency Conversion Fee Antitrust Litigation
(MDL 1409) (Class Action Complaint and Preliminary Approval of Settlement documents
are available at <http://www.ccfsettlement.com/>; on November 8, 2006 the U.S. District
Court for the Southern District of New York approved a class action settlement, by which
Visa and MasterCard agreed to pay $336 million to credit card and debit card holders for
allegedly unlawful currency conversion practices (Visa and MasterCard deny any wrongdo-
ing). The class action suit claimed, among other things, that issuers charged currency con-
version fees that were not appropriately disclosed, violating the provisions of TILA and
EFTA.) On mandatory arbitration clauses—see Dodd–Frank Act, Sec. 1028 (giving the
CFPB authority to ban pre-dispute consumer arbitration); “The Arbitration Trap: How
Credit Cards Companies Ensnare Consumers”, <http://www.citizen.org/publications/
release.cfm?ID=7545>.
2. Fees
The common credit card contract includes a long list of penalty and non-penalty
fees. Non-penalty fees include cash advance, balance transfer, foreign currency
conversion, and expedited-payment fees. These fees apply to optional services
used by some, but certainly not all, cardholders. In the absence of such fees,
issuers would have to cover the cost of these optional services by raising a basic
non-contingent price term—the annual fee, for example. Setting separate fees
or prices for each optional service avoids the cross-subsidization that a single
high annual fee entails and allows for more efficient tailoring of the product to
the needs and preferences of different borrowers.52
Next, consider the penalty fees and rates: the late fee, the over-limit fee,
and the default interest rate. In a rational-choice model, these price terms
reflect efficient risk-based pricing. The assumption is that paying late,
exceeding the credit limit, or engaging in a host of other behaviors that can
trigger a default rate is correlated with an increased risk of non-payment. If
this increased risk is not priced ex post through the penalty fees and rates,
then it would have to be priced ex ante through non-contingent price
terms, such as the general interest rate or the annual fee. The penalty fees
and rates facilitate a better match of risk and price. They also avoid ineffi-
cient cross-subsidization.53
3. A Complex Array of Complex Products

A single credit card contract is complex. And consumers face more than one
contract. When shopping for the credit card that best suits them, consumers
need to read, understand, and compare multiple contracts—a daunting task.
The standard efficiency explanation for the large variety of products
available in many markets is consumer heterogeneity. In the credit card mar-
ket, different borrowers have different preferences and face different con-
straints. A credit card contract that is ideal for one cardholder could be
52. Furletti, above note 16, at 10 (“In lieu of charging all of their customers an annual fee that
subsidized the costs associated with the behaviors of a few, [issuers] began to assess fees directly
on those customers whose card usage behaviors drove costs higher. As issuers started unbun-
dling costs and creating behavior-based fees, fees rebounded and have again become an
important component of issuer revenues (Figure 6).”).
53. Furletti, while noting the benefits of behavior-based and risk-based pricing, acknowledges
that “[a] pricing structure that better allocates issuer’s risk and servicing expenses has likely
come at a cost in the form of a complex and customized product whose pricing is difficult to
summarize.” Furletti, above note 16, at 18.
c re di t card s 77
wrong for another. With more products to choose from, each cardholder, in
theory, is able to choose the credit card and associated credit card contract
that is best for him or her.This explanation, however, assumes that informed
choice is possible, despite the high level of complexity of the choice
problem.
B. Deferred Costs
We’ve seen how the common credit card contract defers costs and acceler-
ates benefits. Can this pricing pattern be explained within a rational-choice
model? Several rational-choice, efficiency-based explanations have been
offered for the relatively high interest rates that issuers charge. As noted
above, the concern about high interest rates has, in large part, dissipated.
More importantly, focusing solely on the interest rate dimension ignores a
large part of the overall picture.
Dagobert Brito and Peter Hartley have argued that high credit card inter-
est rates can be explained by the transaction costs involved in obtaining
credit from alternative, lower interest rate sources, specifically bank loans.54
Transaction costs may well play an important role, but they cannot account
for the observed pricing patterns in the credit card market. One reason is
that, given current technology, it is no longer clear why the cost of provid-
ing a closed-end loan would be higher than the cost of maintaining a credit
card account. Also, while the transaction-costs model may explain why
credit card issuers set a higher overall price, it does not explain why issuers
systematically choose to use interest rates rather than annual or per-
transaction fees to achieve this higher price.
Similarly, it has been argued that high credit card interest rates are needed
to cover other cost elements, such as the cost of building a viable credit card
portfolio, operating expenses (such as rent and salaries), and the cost of serv-
ices beyond lending that the card provides.55 But while these fixed costs—
or, at least, costs that are fixed with respect to lending—can explain
above-marginal cost pricing, they cannot explain why, of all possible dimen-
sions of the credit card price, issuers choose to use high interest rates to
54. Dagobert L. Brito and Peter R. Hartley, “Consumer Rationality and Credit Cards” (1995) 103
J. Pol. Econ’y. 103 (1995), 4000. See also Zywicki, above note 22, at 100.
55. See Evans and Schmalensee, above note 5, at 249–50, 254–5. See also Zywicki, above note 22,
at 120 (operating costs, rather than the cost of funds, are the main component of issuers’
costs).
cover their fixed costs. Why not use annual fees or per-transaction fees?
Other rational-choice models similarly focus on the interest rate dimension
and thus cannot account for the multidimensional pricing patterns identi-
fied in this chapter.56
An alternative explanation that is not confined to the interest rate dimen-
sion relies on the notion of rational ignorance. According to this theory,
consumers do not read their credit card contracts and, therefore, are una-
ware of or indifferent to variations in different provisions of this contract.
But consumers are extremely sensitive to some components of the credit
card contract; specifically, the short-term components. In other words,
rational-choice models allow for imperfect information. Their failure is in
explaining why consumers are informed about certain dimensions of the
credit card contract but not others.
C. Summary
Credit card contracts are complex. They defer costs and accelerate benefits.
The rational-choice theory leaves an explanatory gap, as it explains some—
but not all—of these design features. The behavioral-economics theory,
which we’ll take a look at next, attempts to fill this gap.
IV. A Behavioral-Economics Theory

As explained in Chapter 1, the design of credit card contracts can be
explained as the product of an interaction between consumer psychology
and market forces. The behavioral-economic account is developed below.
I begin with the complexity feature, in Section A. The theory behind
the deferred-cost feature is presented in Section B. Section C introduces
the concept of salience, which is intimately linked to both the complexity
and cost-deferral features. Salience provides a useful conduit between
the imperfect rationality of cardholders and the contract terms that are
designed for them. Finally, in Section D, I briefly explore to what extent
56. For instance, Loretta Mester, using a screening model with collateralized loans and unsecured
credit card loans, explains why credit card interest rates are not sensitive to reductions in the
bank’s cost of funds. See Loretta J. Mester, “Why Are Credit Card Rates Sticky?”, Econ. Th., 4
(1994), 505. Mester, however, does not explain why credit card interest rates exceed the risk-
adjusted cost of funds, nor does she explain the other pricing patterns identified in this chapter.
c re di t card s 79
the behavioral-economics theory is consistent with the market’s reaction

to the CARD Act of 2009.
This Part develops a behavioral-economics theory of credit cards. It also
provides empirical evidence in support of this theory. This evidence takes
two forms. First, observed contract design features that cannot be explained
within the rational-choice framework provide strong, indirect evidence in
support of the behavioral-economics theory. Second, direct evidence—both
survey evidence and evidence of mistakes in product choice and product
use—shows that many cardholders are imperfectly rational. This evidence
lends further support to the behavioral-economics theory. Moreover, the
direct evidence of imperfect rationality proves the importance of the behav-
ioral account also for design features that can be explained within a rational-
choice framework. Even if a certain design feature can be explained while
maintaining the assumption that cardholders are perfectly rational, this
explanation would be unsatisfactory if most cardholders are shown to fall
short of the perfect rationality ideal.
A. Complexity
The typical credit card contract is complex. It specifies numerous interest
rates, fees, and penalties, the magnitude and applicability of which may be
contingent on unknown future events. As explained in Chapter 1, imper-
fectly rational cardholders would find it very difficult to deal with such
complexity. They would ignore certain price dimensions, miscalculate
others, and, as a result, fail to appreciate the total cost of the credit card
product.
Evidence confirms that many cardholders fall far short of the perfect
rationality ideal. A study conducted by the Center for Responsible Lend-
ing found that the majority of borrowers being charged penalty rates do
not realize it. Consumer testing commissioned by the Federal Reserve
Board (FRB) revealed that (1) many credit card fees go unnoticed, (2)
even consumers who are aware of the different credit card rates and fees
do not understand how those rates and fees would be applied, and (3)
consumers do not notice the cumulative effect of paying small amounts of
fees every month. A study by Sumit Agarwal, John Driscoll, Xavier Gabaix,
and David Laibson found that consumers also do not understand the
workings of balance-transfer offers. Another study by the Center for
Responsible Lending found that only 3 percent of borrowers have the
knowledge and capacity to evaluate credit card companies’ payment-

allocation policies.57
On a more general level, a survey conducted by the Center for American
Progress found that 38 percent of consumers believe that “[m]ost financial
products, such as mortgage loans and credit cards, are too complicated and
lengthy for [them] to fully understand.” A report commissioned by the FRB
noted that a significant number of consumers “lack fundamental under-
standing of how credit card accounts work.” And the General Accounting
Office, in a report to Congress, concluded that credit card contracts are too
complex for consumers to understand.58
Increased complexity may be attractive to issuers, as it allows them to
hide the true cost of the credit card in a multidimensional pricing maze.
An issuer who understands the imperfectly rational response to complex-
ity can use complexity to its advantage by creating an appearance of a
lower total price without actually lowering the price. For example, if the
currency-conversion fee and the cash-advance fee are not salient to card-
holders, issuers will raise the magnitude of these price dimensions.59
Increasing these prices will not hurt demand. On the contrary, it will
enable the issuer to attract cardholders by reducing more salient price
dimensions or by increasing more salient benefit dimensions (like reward
points and frequent-flyer miles). This strategy depends on the existence of
non-salient price dimensions. When the number of price dimensions goes
up, the number of non-salient price dimensions can also be expected to
go up. Issuers thus have a strong incentive to increase complexity and
multidimensionality.
57. See Center for Responsible Lending, “Priceless or Just Expensive? The Use of Penalty Rates
in the Credit Card Industry,” December 16, 2008; Macro International, “Design and Testing
of Effective Truth in Lending Disclosures” (2007) at vii, 9, 26, available at <http://www
.federalreserve.gov/dcca/regulationz/20070523/Execsummary.pdf>; Ann Kjos, “Proposed
Changes to Regulation Z: Highlighting Behaviors that Affect Credit Costs” (2008) FRB of
Philadelphia, Payment Cards Center Discussion Paper No. 08–02, available at SSRN: <http://
ssrn.com/abstract=1160257>; Sumit Agarwal, John Driscoll, Xavier Gabaix, and David Laib-
son, “The Age of Reason: Financial Decisions over the Life-Cycle and Implications for Reg-
ulation” (2009) Brookings Papers on Economic Activity. Issue 2, 51–117; Center for Responsible
Lending, “What’s Draining Your Wallet? The Real Cost of Credit Card Cash Advances,”
December 16, 2008.
58. Ctr. for Am. Progress et al., Frequency Questionnaire (2006), <http://www
.americanprogress.org/kf/debt_survey_frequency_questionnaire.pdf>; Macro International,
above, at 91; GAO Complexity Report, above note 47, at 49.
59. See Furletti, above note 16, at 12 (noting that service fees, which are likely non-salient, “are
generally priced to provide attractive profit margins.”).
c re di t card s 81
Indeed, there is evidence that the high level of complexity was a deliber-
ate design feature of the credit card contract. Shailesh Mehta, former CEO
of Providian, acknowledged that credit card pricing was designed so that it
would require “some kind of degree” to understand.60
Complexity can be expected to increase as cardholders learn to effect-
ively incorporate more price dimensions into their decision-making. If issu-
ers significantly increase the magnitude of a non-salient price dimension,
cardholders will eventually learn to focus on this price dimension and it will
become salient. Issuers will have to find another non-salient price dimen-
sion. When they run out of non-salient prices in the existing contractual
design, they can create new ones by adding more interest rates, fees, or
penalties.
B. Deferred Costs
1. General
The behavioral-economics explanation for deferred-cost contracts is based
on evidence that future costs are often underestimated. When future costs
are underestimated, contracts with deferred-cost features become more
attractive to cardholders and thus to issuers. Consider a simplified credit
card contract with two price dimensions: a short-term price, PST , such as an
introductory interest rate, and a long-term price, PLT , such as a long-term
interest rate. Assume that the optimal credit card contract sets PST = 0.1 and
PLT = 0.1, as these prices provide optimal incentives and minimize total
costs. If cardholders are rational, issuers will offer this optimal contract.
Now assume that cardholders underestimate future costs. For example,
assume that cardholders underestimate the likelihood of borrowing on their
credit card after the introductory period: while they will borrow an amount
of $100 both during and after the introductory period, when they obtained
the credit card they predicted that they would borrow $100 during the
introductory period but only $50 after the introductory period ends.
As a result of such misperception, issuers will no longer offer the optimal
contract. To see this, compare the optimal contract, the (0.1,0.1) contract,
with an inefficient, deferred-cost contract setting PST = 0.05 and PLT = 0.16,
the (0.05,0.16) contract. Assume that under both contracts, the issuer just
covers the total cost of offering the credit card product; under the optimal
60. See Interview in Frontline, “The Credit Card Game,” November 24, 2009.
(0.1,0.1) contract, total interest payments are: P(0.1,0.1) = 0.1 · 100 +

0.1 · 100 = 20 (assuming, for clarity of exposition, that the introductory
period and the post-introductory period are one year each and that interest
is assessed at the end of the period; time-discounting is also ignored for sim-
plicity). Under the inefficient (0.05,0.16) contract, total interest payments are:
P(0.05,0.16) = 0.05 · 100 + 0.16 · 100 = 21. The total-cost and total interest
payments are higher under the inefficient, deferred-cost contract.
Now consider the cost of the credit card as perceived by the imperfectly
rational cardholder. Perceived total interest payments under the optimal
(0.1,0.1) contract are: P̂(0.1,0.1) = 0.1 · 100 + 0.1 · 50 = 15. Perceived total
interest payments under the inefficient (0.05,0.16) contract are: P̂(0.05,0.16)
= 0.05 · 100 + 0.16 · 50 = 13. Cardholders would prefer, and thus lenders
will offer, the inefficient, deferred-cost contract.
We’ve seen that if future costs are underestimated, issuers will offer
deferred-cost contracts. Let’s now consider the idea that cardholders under-
estimate future costs. Two underlying biases are responsible for this underes-
timation; myopia and optimism. A myopic cardholder focuses on short-term
benefits and ignores or discounts long-term costs. An optimistic cardholder
underestimates self-control problems with using the card and the likelihood
of unexpected developments that may bring economic hardship. This opti-
mism results in underestimation of future borrowing.61 Since many long-
term price dimensions in the credit card contract are contingent upon
borrowing, the underestimation of future borrowing leads to an underesti-
mation of future costs.
The effect of myopia on the underestimation of future costs is obvious.
The effect of optimism is more subtle. As noted above, in the credit card
market, optimism operates on perceptions of the extent of self-control and
of the likelihood of contingencies bearing economic hardship. These mani-
festations of the optimism bias are considered in subsections 2 and 3 below.
The cost of credit card debt depends not only on how much is borrowed,
but also on how fast the debt is paid off. Optimism also inflicts predictions
about repayment speed, as explained in subsection 4. Other misperceptions
are discussed in subsection 5.
61. See Ausubel, above note 11, at 70–1 (“[T]here are consumers who do not intend to borrow
but continuously do so”); Lawrence M. Ausubel, “Adverse Selection in the Credit Card
Market” (1999) 20 (unpublished manuscript) (empirically testing and confirming the “under-
estimation hypothesis”).
c re di t card s 83
2. Optimism Take 1: Underestimating Self-Control Problems

Imperfect self-control provides one major explanation for consumers’
underestimation of their future borrowing. Many consumers overestimate
their ability to resist the temptation to finance consumption by borrowing;
consequently, they underestimate future borrowing.62
When obtaining a credit card, the consumer may intend to use the card for
transacting only or to limit borrowing to a certain amount. But this limit is not
specified in the credit card contract, and therefore is not binding on the con-
sumer’s “future self,” the self that will make the borrowing decision. And with
imperfect self-control, this future self may well exceed the intended limit.
a. Hyperbolic Discounting
Why would consumers end up borrowing more than they initially antici-
pated? What is the source of such weakness of the will? The answer can be
traced to a concept called “hyperbolic discounting.” Neoclassical economics
assumes that individuals discount the future at a constant rate, an assumption
captured by an exponential discount function. In contrast, consumers are
said to be a hyperbolic discounters if their short-run discount rate is larger
than their long-run discount rate. In other words, at a given point in time, t,
a hyperbolic discounter heavily discounts costs and benefits that will materi-
alize in the near future, at t + 1, but assigns only a smaller additional discount
for costs and benefits that will materialize in the more distant future, at t + 2.
This systematic disparity between people’s short-term and long-term dis-
count rates has been consistently demonstrated both in the laboratory and
in the real world.When hyperbolic discounters are naive about the nature of
their time preferences, they will optimistically overestimate their willpower,
and consequently underestimate their future borrowing.63
62. See Evans and Schmalensee, above note 5, at 109 (“[R]eal people have trouble keeping track
of [card] balances during the month and resisting the ever-present temptation to use future
income to enjoy life a bit more today.”); Thomas H. Jackson, The Logic and Limits of Bankruptcy
Law (Harvard University Press, 1986) 234, 238–99 (hereinafter Logic and Limits) (arguing that
individuals underestimate “the risks that their current consumption imposes on their future
well-being,” and invoking the “human tendency to lack impulse control”—that is, a tendency
“to choose current over postponed gratification, even if it is known that the latter holds in
store a greater measure of benefits”).
63. On exponential discounting in neoclassical economics—see Paul Samuelson, “A Note on
Measurement of Utility”, Rev. Econ. Stud., 4 (1937), 155; Tjalling C. Koopmans, “Stationary
Ordinal Utility and Impatience”, Econometrica, 28 (1960), 287. For experimental evidence sup-
porting the hyperbolic discounting model—see, e.g., Richard H. Thaler, “Some Empirical
Evidence on Dynamic Inconsistency”, Econ. Letters, 8 (1981), 201, 202 (one of the first
experiments); Shane Frederick et al., “Time Discounting and Time Preference: A Critical
Consider a consumer who decides to obtain a credit card at T = 0. From

the T = 0 perspective, this consumer considers the likelihood of borrowing
on the credit card at T = 1. The consumer weighs the future benefit of a
credit card purchase at T = 1 against the more distant T = 2 cost of debt
repayment, including payment of interest charges. Recall that from the T = 0
perspective, the discount between T = 1 and T = 2 is relatively small. Hence,
given the substantial costs of credit card borrowing, even though the costs
of borrowing lie in the more distant future, at T = 0 the consumer would
prefer not to borrow at T = 1. And assuming the consumer thinks that this
ex ante preference will be followed, at T = 0, the consumer believes that no
borrowing will take place at T = 1.
To examine the validity of this belief, let’s move down the timeline to
T = 1, when the actual borrowing decision takes place. At this point, T = 1
is the present and T = 2 is the near future. As explained above, hyperbolic
discounting implies that from the T = 1 perspective the T = 2 costs will be
heavily discounted. Therefore, even if the future (T = 2) cost of borrowing
is substantially higher than the present (T = 1) benefits, the consumer, at
T = 1, may decide to borrow on his or her credit card.
This preference reversal—a T = 0 preference not to borrow evolving into
a preference and a decision to borrow at T = 1—is an immediate implica-
tion of hyperbolic discounting. Figure 2.1 illustrates the reversal of prefer-
ences with respect to credit card borrowing.
Review”, J. Econ. Lit., 40 (2002), 351, 360 (a recent survey). Outside the laboratory, hyperbolic
discounting is evident in consumption decisions;—see Richard H. Thaler, “The Winner’s Curse:
Paradoxes and Anomalies of Economic Life” (Princeton University Press, 1992) 94, 105; insufficient
saving for retirement—see Brigitte C. Madrian and Dennis Shea, “The Power of Suggestion:
Inertia in 401(k) Participation and Savings Behavior”, Q. J. Econ., 116 (2001), 1149, 1150; David
Laibson et al., “Self-Control and Saving for Retirement” (1998) 1 Brookings Papers on Economic
Activity 91; Ted O’Donoghue and Matthew Rabin, “Procrastination in Preparing for Retire-
ment,” in Henry Aaron (ed.), Behavioral Dimensions of Retirement Economics (Brookings Institute
Press, 1999) 125; addiction, whether to cigarettes, alcohol, or more serious drugs—see George
Ainslie,“Derivation of ‘Rational’ Economic Behavior from Hyperbolic Discount Curves” Amer.
Econ. Rev. 81 (1991) 334; Jonathan Gruber and Botond Koszegi,“Is Addiction ‘Rational’? Theory
and Evidence” (2000) NBER Working Paper No. 7507; dieting—see Thaler, The Winner’s Curse,
id., at 98; and health club attendance—see Stefano DellaVigna and Ulrike Malmendier, “Paying
Not to Go to the Gym”, Amer. Econ. Rev., 96 (2006), 694. In a recent study, Laibson et al. show
that allowing for hyperbolic discounting helps explain the large fraction of households that bor-
row on their credit cards. See Laibson et al., above note 40, at 229–30. See also George-Marios
Angeletos et al.,“The Hyperbolic Consumption Model: Calibration, Simulation, and Empirical
Evaluation” J. Econ. Perspect. 15 (2001) 47 (“[H]ouseholds with hyperbolic discount functions are
very likely to borrow on their credit cards to fund instant gratification. Thus households with
hyperbolic discount functions are likely to have a high level of revolving debt, despite the high
cost of credit card borrowing.”).
c re di t card s 85
Cost of
borrowing
Benefit
from
borrowing
T
0 1 2
Figure 2.1. Preference reversal in credit card borrowing
The vertical line at T = 1 represents the T = 1 value of the benefits from

borrowing. The curved line descending from this vertical line (to the left)
represents the discounted value of these benefits at any point in time prior
to T = 1, and especially at T = 0. Similarly, the vertical line at T = 2 represents
the T = 2 value of the costs associated with credit card borrowing. And the
curved line descending from this vertical line (to the left) represents the dis-
counted value of these costs at any point in time prior to T = 2, and espe-
cially at T = 1 and at T = 0. As illustrated in Figure 2.1, the two curved lines
start with a steep descent, which then levels off as the temporal distance from
the curve’s point of origin increases. This varying slope of the discounted
present value curves captures the hyperbolic discounting phenomenon.
In Figure 2.1 we see that at T = 0, the curve representing the discounted
present value of the costs of borrowing lies above the curve representing the
discounted present value of the benefits from borrowing. Hence, at T = 0, the
consumer would prefer not to borrow (at T = 1). However, between T = 0
and T = 1 the relative position of the two curves switches, and at T = 1 the
benefits curve lies above the costs curve, implying that the consumer will,
in fact, choose to borrow at T = 1.
Figure 2.1 illustrates the temporal inconsistency resulting from hyper-
bolic discounting. At T = 0, the consumer does not wish to borrow, but ends
up doing so anyway at T = 1. Note, however, that temporally inconsistent
preferences do not necessarily entail inaccurate ex ante beliefs. Sophisti-
cated consumers, aware of their hyperbolic discounting, would anticipate
the preference reversal. At T = 0, such consumers, while preferring not to
borrow at T = 1, would nevertheless know that they will end up borrowing

at T = 1. Unfortunately, not many consumers are that sophisticated with
respect to their intertemporal preferences. Many consumers are at least a
little naive, at T = 0, about their ability to effectuate, at T = 1, their T = 0
preferences. Specifically, they might fail to take into account the T = 1 pref-
erence reversal when making the T = 0 decision.64 Naive hyperbolic dis-
counters thus believe ex ante that they will not borrow, but actually do
borrow on their credit card ex post. In other words, the hyperbolic dis-
counter optimistically underestimates future borrowing.
Note how credit cards, as an open-end credit vehicle, give rise to
imperfect self-control and to the underestimation of future borrowing.
When a consumer takes on a closed-end loan, all the parameters of the
loan contract, including the amount of the loan, are determined up-front.
No discretion is reserved for a later period; thus, self-control is not an
issue. The credit card, on the other hand, separates the decision to obtain
a card—and the decision which card to obtain—from the actual borrow-
ing decision (or decisions). The amount of the loan is left open. And an
open-end loan inevitably opens the door to self-control problems. In
other words, a closed-end loan serves as a commitment device, enabling
the consumer to constrain his or her future self by committing to a maxi-
mum amount of debt. The credit card does not provide such a commit-
ment device.65
A recent study by Stephan Meier and Charles Sprenger directly tests, and
confirms, the behavioral-economics account. Meier and Sprenger compare
time-preference data from a field experiment with a “targeted group of
low-to-moderate income consumers,” with credit report data on these con-
sumers. The authors find that consumers who exhibit hyperbolic discount-
ing and dynamically inconsistent intertemporal choices borrow more, and
specifically borrow more on their credit cards. This result suggests that
64. See, e.g., Madrian and Shea, above note 63 (evidence from 401(k) investments); DellaVigna
and Malmendier, above note 63 (evidence from health club attendance).
65. Many accounts of credit card borrowing mention consumers’ weakness of will, or imper-
fect self-control. The notion is that consumers do not intend to borrow, but end up doing
so anyway. See, e.g., Ausubel, above note 10, at 262. The credit card borrower is sometimes
compared to an alcoholic, who is aware of the dangers inherent in her drinking problem,
yet cannot avoid purchasing another bottle. Another common analogy compares the credit
card borrower to a failed dieter. The dieter wants to lose weight, but when that chocolate
cake presents itself, he cannot resist the temptation. The alcoholism and dieting analogies
highlight the self-control problem. See Sullivan, Warren, and Westbrook, above note 1, at
120, 247, 250.
c re di t card s 87
“individuals borrow more . . . than they actually would prefer to borrow

given their long-term objectives.”66
b. Borrowing a Little at a Time
Imperfect self-control leading to the underestimation of future borrowing
has been traced back to the temporal separation between the decision to
obtain a credit card and the decision to borrow on the credit card. The fact
is, however, that there are many borrowing decisions, not just one. Each
time consumers swipe their card, they take on a new loan. This piecemeal
borrowing phenomenon, or “a-little-at-a-time borrowing,” exacerbates the
self-control problem.67
A 2008 Dēmos survey found that smaller purchases of non-essential goods
and services, such as meals at restaurants, movies, and DVDs, were the most
frequently cited expense category contributing to credit card debt (cited by
48 percent of respondents). Sullivan, Warren, and Westbrook observe that
“[d]ebtors who never dream of seeking a $5,000 bank loan might run up
$5,000 in charges of $50 at a time.” The distinction between the traditional
discrete loan and the gradually accumulating credit card debt should not be
underestimated: “One need not have a deep understanding of human nature
to appreciate the risks of incremental foolishness. There are many mistakes
we would not make all at once that we will make a little at a time. . . . ” 68
As fallible decision-makers, we inevitably make mistakes. But we will try
harder to avoid such mistakes—wisely so—when the stakes are higher (or
appear to be higher). This general observation is applicable to decisions
about incurring additional debt: “The debt itself is incurred a little bit at a
time, so that even large amounts of debt do not involve a single, sober deci-
sion to take on $25,000 or even $2,500 of debt.”69
Credit cards opened the door to “the seductiveness of incremental irre-
sponsibility,” as manifested in a-little-at-a-time borrowing. The outcome
of such borrowing behavior is usually detrimental to consumers. Sullivan,
66. Stephan Meier, and Charles Sprenger. “Present-Biased Preferences and Credit Card Borrow-
ing”, Amer. Econ. J., 2 (2010), 193.
67. See Sullivan, Warren, and Westbrook, above note 1, at 130 (“[C]redit cards make it far easier
to incur consumer debt by encouraging a-little-at-a-time borrowing and too-little-at-a-time
repayment.”).
68. See Dēmos, “The Plastic Safety Net: How Households Are Coping in a Fragile Economy”
(2009) 6;Teresa A. Sullivan, Elizabeth Warren, and Lay Lawrence Westbrook, As We Forgive Our
Debtors: Bankruptcy and Consumer Credit in America (Oxford Universtiy Press, 1989) 178;
Sullivan, Warren, and Westbrook, above note 1, at 247.
69. Sullivan, Warren, and Westbrook, above note 1, at 245–6.
Warren, and Westbrook describe a category of debtors, whom they call

“sliders”: “[m]any people slide into debt, falling a little farther behind on
their cards every month until bankruptcy is the only way out.” Optimism
prevents cardholders from fully appreciating the risks of a-little-at-a-time
borrowing, resulting in the underestimation of future borrowing and of
deferred costs associated with borrowing.70
3. Optimism Take 2: Underestimating the Risk of Economic Hardship

Evidence shows that a substantial amount of credit card borrowing is trig-
gered by adverse contingencies, especially job loss and medical problems,
which decrease income and increase expenses, respectively.71 Underestima-
tion of future borrowing may stem from an optimism bias that leads con-
sumers to underestimate the likelihood of adverse events that might generate
a need to borrow. Optimistic individuals tend to underestimate the proba-
bility that they or a loved one will have an accident or illness that requires
costly treatment (not covered, or not entirely covered, by insurance). Indi-
viduals also tend to underestimate the likelihood that they will lose their
job, or underestimate the length of time it will take to find a new job.72
These and other manifestations of the optimism bias will lead consumers to
underestimate the likelihood that they will be forced to resort to credit card
borrowing.
The underestimation of future borrowing has been traced back to two
manifestations of the optimism bias—a consumer’s overestimation of will-
power and underestimation of the likelihood of encountering adverse con-
tingencies. The two problems may reinforce one another. Imperfect will-
power might push the consumer into a fragile financial condition, increasing
vulnerability to adverse events, such as an accident, illness, or job loss.73
70. See Sullivan, Warren, and Westbrook, above note 68, at 179; Sullivan, Warren, and Westbrook,
above note 1, at 111.
71. See Dēmos, “The Plastic Safety Net: The Reality behind Debt in America” (2005) 7–12,
available at <http://www.accessproject.org/adobe/the_plastic_safety_net.pdf>; Dēmos, above
note 68, at 8 (“[M]ore than one-half of indebted low- and middle-income households cited
medical expenses as contributing to their credit card debt. In fact, compared to all other
expenses inquired about in the survey, out-of-pocket medical expenses was the most fre-
quently reported expense that contributed to credit card debt. On average, these households
reported that $2,194 in credit card debt was attributable to medical expenses.”).
72. Sullivan,Warren, and Westbrook, above note 1, at 25, 114 (“The recently unemployed, hopeful
that they will be back at work in a matter of days or weeks, may not be prepared to tell the
children there will be no new soccer shoes this season or no back-to-school clothes.”).
73. See Sullivan, Warren, and Westbrook, above note 1, at 113–15 (Many consumers reach a state
of indebtedness that renders them vulnerable to unexpected costs). See also Bruce A. Markell,
c re di t card s 89
4. Optimism Take 3: Underestimating the Cost of Slow Repayment

The amount to be borrowed is clearly a major factor affecting the cost of
borrowing. But it is not the only factor.The cost of borrowing also depends
on the speed of repayment. Many borrowers underestimate the period it
will take them to repay their credit card debt. The hyperbolic discounting
phenomenon, which accounts for the underestimation of future borrowing,
also explains consumers’ underestimation of the repayment period. Naive
hyperbolic discounters often anticipate a quick repayment schedule, but
when actual payments need to be made revert to the minimum payment.
According to one account: “Each month the debtor might make the small
minimum payment with a vow to start paying off the balance the next
month.” Again, optimism about self-control problems results in the under-
estimation of the costs associated with credit card debt.74
Optimistic underestimation of the risk of economic hardship can simi-
larly lead borrowers to overestimate the speed of repayment and thus under-
estimate the costs associated with credit card debt. Optimistic cardholders
often overestimate their ability to repay quickly because they underestimate
the likelihood of adverse contingencies that would render quick repayment
difficult. Evidence of low repayment rates is consistent with this behavioral
explanation. In 2008, the monthly payment rate—the amount that card-
holders pay on their credit card debt—hovered around 18 percent.75 The
same factors that lead consumers to underestimate future borrowing will
lead them to also overestimate their ability to repay quickly.
Because issuers profit from slow repayment, they often set low minimum
payments.They even design the credit card bill such that the minimum pay-
ment figure is more salient than the total balance figure. They do this to
“persuade” consumers to pay only the minimum payment. For instance, the
“minimum payment” box is often closer to the “actual payment” box and
highlighted with a distinct color or font size, while the “total balance” box
“Sorting and Sifting Fact from Fiction: Empirical Research and the Face of Bankruptcy: The
Fragile Middle Class: Americans in Debt By Teresa A. Sullivan, Elizabeth Warren and Jay
Lawrence Westbrook” (2001) 75 Am. Bankr. L.J. 149 (book review) (“Americans who file
bankruptcy generally have incurred debt beyond a rational ability to repay, and are thus vul-
nerable to economic events that challenge their fragile condition.”).
74. See Sullivan, Warren, and Westbrook, above note 68, at 178; Paul Heidhues and Botond
Koszegi, “Exploiting Naiveté about Self-Control in the Credit Market”, Amer. Econ. Rev. 100
(2010), 2279.
75. CardFlash, “Payment Rates Hit a Four-Year Low in July,” August 27, 2008; CardFlash,
“Monthly Payment Rates Edge Up Slightly,” March 18, 2009.
is further away and less conspicuous. Sometimes the minimum payment

figure is the only figure appearing on the payment stub itself.76
5. Other Misperceptions
Deferred-cost contracts are a response to the underestimation of future
costs. We’ve seen how imperfect rationality can lead cardholders to under-
estimate future costs by underestimating future borrowing or overestimat-
ing the speed of repayment. But future costs can also be underestimated
without regard to the extent of borrowing or the speed of repayment.
For instance, the common credit card contract sets high penalty fees for
late payments and for exceeding the credit limit. These design features can
be linked to the misperceptions discussed above; optimism about the strength
of self-control or about the risk of financial hardship. Imperfect self-control
may result in higher-than-expected borrowing, perhaps even above the
consumer’s credit limit, triggering over-limit fees. High debt levels imply
higher minimum payments and thus might lead to late payment, triggering
late fees. Also, consumers may be forced to exceed their credit line or to
defer payment beyond the due date as a result of an accident, an illness, or
unemployment, thus linking over-limit fees and late fees to the underesti-
mation of adverse contingencies.
But consumers might also miss the due date simply through forgetfulness.
Similarly, they might exceed their credit line simply because they lost track
of the total balance. Since forgetfulness is a common trigger of the penalty
clauses in the credit card contract, optimism regarding the extent of such for-
getfulness is important in explaining these features of credit card pricing.77
76. See Scott D. Schuh and Joanna Stavins, “Summary of the Workshop on Consumer Behavior
and Payment Choice” (2008) FRB of Boston Public Policy Discussion Paper No. 08–5, 9, avail-
able at SSRN: <http://ssrn.com/abstract=1310350> (“monthly credit card statements
emphasize the minimum payment due—often denoted in bold font—but not the total
amount of debt.”) See also CardFlash, “Minimum Payments,” December 17, 2008. (A 2008
study by Neil Stewart found that the average monthly payment on a credit card account
increased by 70 percent when the minimum monthly payment amount was not included on
the statement. The study suggests that consumers are encouraged to pay less on their bal-
ances when credit card companies present the minimum monthly payment option amount
on credit card statements.)
77. Consistent with the forgetfulness explanation, Satngo and Zinman find that many consumers
could have easily avoided paying late fees and over-limit fees by either paying a bill using
available checking balances, or by using a different card with sufficient available credit. See
Victor Stango and Jonathan Zinman, “What Do Consumers Really Pay on Their Checking
and Credit Card Accounts? Explicit, Implicit, and Avoidable Costs”, Amer. Econ. Rev. 99
(2009), 424–9.
c re di t card s 91
Other features of the credit card contract may be linked to different biases
and misperceptions. A consumer shopping for a credit card may underesti-
mate the need to use the credit card abroad, and thus underestimate the
likelihood of paying a currency-conversion fee. The cardholder may also
underestimate the benefit of paying the credit card bill over the phone and
thus underestimate the likelihood of incurring a “convenience” fee.
The underlying source of these underestimation biases is not always clear.
Perhaps the temporal dimension is once again playing a role: individuals
excessively discount the likelihood of future events, such as traveling abroad
or the number of times when paying by telephone may be especially con-
venient. When the future is difficult to imagine, the probability of future
events even taking place will be discounted.
Consumers might pay insufficient attention to certain terms, such as a
currency-conversion or convenience fee, because they underestimate the
probability of triggering these fees. They may also pay insufficient attention
to these fees simply because such fees never cross their minds. The signifi-
cance of a contract term can be underestimated because the probability of
triggering the term is perceived to be small. But a term can also be ignored
for reasons that are more difficult to identify and articulate.
C. Salience
Faced with the complex, multidimensional credit card contract, imperfectly
rational consumers will not be able to focus equally on all terms. Only a
handful of terms will be salient. Salience is, therefore, inexorably linked to
the complexity feature. The preceding analysis has shown that salience has
an important temporal dimension. Short-term prices are generally more
salient than long-term prices. This explains the deferred-cost feature. But
while salience has an important temporal dimension, timing is not the only
factor affecting the salience of a price or other contractual dimension. For
example, advertising by issuers or government-mandated disclosures can
make a long-term price salient to consumers.
In any event, an issuer operating in a competitive market will quickly
discover which features are salient and which are not. Salient features will
be made attractive by lowering prices on those features and increasing the
benefits that they provide. Non-salient features, on the other hand, will
constitute the revenue centers. They will be designed to cover the issuer’s
costs and pay for the salient benefits.
1. Salience and the Design of Credit Card Contracts

The behavioral-economics theory explains the design considerations behind
many of the features of the common credit card contract. Many long-term
price dimensions are less salient to cardholders.These include late fees, over-
limit fees, cash-advance fees, and currency-conversion fees. These fees are
set at a relatively high level. Many short-term dimensions are more salient
to cardholders. These include annual fees and introductory interest rates.
Both are set at very low levels, with a zero annual fee and a zero teaser rate
being quite common.78
The notion that short-term dimensions are more salient than long-term
dimensions has a strong empirical basis. Evans and Schmalensee, in their
book, cite evidence that a zero annual fee is the prime selection criterion in
credit card choice for 15 percent of consumers. Similarly, Lawrence Ausubel
notes: “the experience of credit card marketers is that consumers are much
more sensitive to increases in the annual fee than to commensurate increases
in the interest rate.”79
Regarding teaser rates, despite the fact that most borrowing is done at high
post-promotion rates, consumers appear to be extremely sensitive to teaser
rates. Ausubel found that “consumers are at least three times as responsive to
changes in the introductory interest rate as compared to dollar-equivalent
changes in the post-introductory interest rate.”80 Survey evidence suggests
that more than a third of all consumers consider an attractive introductory
interest rate to be the prime selection criterion in credit card choice.81
Scrutinizing consumers’ choices among competing teaser-rate offers
lends further support to the behavioral-economics model: Shui and Ausubel
found that when faced with otherwise identical credit card offers, consum-
ers prefer a credit card with a 4.9 percent teaser rate lasting for an introduc-
tory period of six months over a credit card with a 7.9 percent teaser rate
lasting for an introductory period of twelve months. Consumers in this
study carried an average balance of $2,500 over a one-year period. Those
who accepted the six-month introductory offer paid a post-introductory
rate of 16 percent during the latter half of the year. These results indicate
78. The theory behind teaser rates is a bit more subtle. See Appendix.
79. See Evans and Schmalensee, above note 5, at 225; Ausubel, above note 11, at 72.
80. See Ausubel, above note 61, at 21. Moreover, “consumers are two to three times as responsive
to changes in the introductory interest rate as compared to dollar-equivalent changes in the
duration of the introductory offer.” Id. at 22.
81. See Ausubel, above note 61, at 21; Evans and Schmalensee, above note 5, at 225.
c re di t card s 93
that at least some consumers were making a significant mistake, opting for
the lower-rate, shorter-duration card even though they paid $50 more in
interest on this card than they would have with the longer-duration alterna-
tive. One possible explanation: Consumers systematically underestimate the
amount that they will borrow, or at least the amount they will borrow on a
specific card, in the post-introductory period.82
As noted above, the correlation between temporal distance and salience
is not perfect. Transacting, like borrowing, occurs in the future, relative to
the time period in which the cardholder chooses a credit card and a credit
card contract.Yet, there are no per-transaction fees; in fact, once the benefits
or rewards commonly attached to credit card purchases are taken into
account, many cardholders enjoy a negative per-transaction fee.83
Of course, transacting and borrowing are very different. The optimism-
related biases explaining the underestimation of future borrowing do not
apply to transacting. Consumers applying for a credit card, while hoping that
they will not need to borrow on their card, surely expect to use the card for
transacting purposes. Accordingly, transacting and transacting-related pricing
are salient. Many cardholders rank loyalty programs among the top features
they look for when choosing a credit card.84 As previously noted, the inter-
change fee, to the extent that it translates into higher purchase prices, can be
viewed as a per-transaction fee. But this only reinforces the behavioral-
economics theory because when charging a per-transaction fee, issuers make
sure to do so in an indirect way that reduces its salience to consumers.
Similarly, the basic (not introductory) interest rate on purchases is a long-
term price dimension, which is nonetheless quite salient to cardholders and
thus the subject of competition among issuers. It is noteworthy that this was
not always the case. As explained above, until fairly recently the basic interest
rate was not salient to cardholders, and thus the interest rate was quite high.
But salience is a dynamic concept. Over time, consumers learned to appre-
ciate the importance of the interest-rate dimension, arguably with some
assistance from legislators and regulators.When interest rates became salient
82. See Shui and Ausubel, above note 28, at 8–9.

83. There is some evidence that rewards come with higher long-term rates, consistent with the
behavioral-economics theory. See CardFlash, “Reward Card Review,” June 2, 2008 (accord-
ing to Consumer Reports, cash-back, gas, and grocery rewards credit cards can offer some
relief for costly essential items, but often carry higher rates than traditional credit cards).
84. Evans and Schmalensee cite survey evidence suggesting that 20 percent of consumers con-
sider rewards and rebates to be the prime selection criterion in their credit card choice. See
Evans and Schmalensee, above note 5, at 225.
to consumers, competition focused on this price dimension, as the

behavioral-economics theory predicts.
Perhaps the most convincing evidence supporting the behavioral-eco-
nomics theory comes from “admissions” by the credit card industry: David
Evans and Richard Schmalensee, the Visa consultants, acknowledge that
“[s]ervice fees (such as late fees, over-limit fees, and finance charges on cash
advances) provide revenues to issuers but are likely to be largely invisible to
most consumers trying to choose between different credit card plans.” Evans
and Schmalensee also acknowledge that issuers reduce salient fees while
increasing non-salient fees: “It is certainly possible, in theory, that issuers who
reduce their annual fees may raise other fees less visible to consumers. Avail-
able data on annual and service fees [e.g. late fees, over-limit fees] strongly
suggest that, over time, issuers in the aggregate have done just this.”85
Similarly, in Beasley v. Wells Fargo Bank, the bank’s “Credit Card Task
Force” proposed increasing “late” and “over-limit” fees as a “good source of
revenue.” Penalty fees are considered a “good source of revenue,” because
the industry perceives—in line with the behavioral-economics theory
developed here—that “[t]here (are) very few cardholders that switch cards
because the late fee is too high.” As explained by Shailesh Mehta, former
CEO of Providian, issuers increased late fees, because people do not think
they will be late. If people do not think they will be late, they will not
switch cards because the late fee is too high.86
Industry sources also confirm the behavioral-economics prediction that
issuers will aggressively compete on salient price dimensions and recoup
losses through non-salient price dimensions. Andrew Kahr, a financial serv-
ices consultant who is credited with inventing the zero-percent teaser rate
(among other consumer-lending strategies), argued that issuers use penalty
fees and rates to recoup losses on teaser rates.87
2. Dynamics of Salience
The behavioral-economics theory predicts low prices on salient price
dimensions and high prices on non-salient price dimensions. Salience thus
85. See Evans and Schmalensee, above note 5, at 211, 260.

86. See Beasley v.Wells Fargo Bank, 235 Cal. App. 3d 1383, 1389 (1991);“Credit Card Fees Soar Again,”
CNNMoney, August 18, 1998 (quoting Peter Davidson, Executive VP at Speer and Associates in
Atlanta, on the industry’s belief that high late fees will not drive cardholders away); Frontline,
“The Credit Card Game,” November 24, 2009 (interview with Shailesh Mehta).
87. See Patrick McGeehan, “Soaring Interest Compounds Credit Card Pain for Millions,” New
York Times, November 21, 2004 (quoting Mr. Kahr).
c re di t card s 95
becomes critical to the analysis of market outcomes. As explained earlier,

salience has a strong temporal dimension, which explains the cost-deferral
feature in many credit card contracts. But salience has other non-temporal
dimensions as well. Moreover, salience is fluid, evolving over time. A non-
salient price or term can eventually become salient.
For example, before the early 1990s, the annual fee was salient to con-
sumers and issuers competed by lowering or waiving the annual fee. At the
time, the interest rate—the basic interest rate for purchases—was not salient
to consumers. Accordingly, issuers did not compete on interest rates. This
changed in the early 1990s: Consumer awareness of the purchase Annual
Percentage Rate (APR) increased and interest rates decreased.The increased
salience of the APR is probably responsible, at least in part, for the teaser rate
innovation as well.88 (Other forces contributing to the success of teaser rates
include consumer myopia and optimism about the extent of borrowing
beyond the introductory period and the ability to switch to another card
with a new teaser rate.)
As competition focused on interest rates, interest revenues declined, and
issuers set out to find, or create, new non-salient price dimensions. Starting in
the late 1990s, issuers shifted focus to behavior-based fees, such as risk-based
fees and convenience fees. Late fees, over-limit fees, and cash-advance fees
increased significantly and new fees—such as returned-check (NSF) fees,
balance-transfer fees, foreign currency exchange surcharges, and expedited-
payment fees—were introduced. According to one industry source, over-limit
fees did not exist in the 1980s and late fees averaged about $10, reflecting the
cost of making a reminder phone call or sending a reminder letter. In the late
1990s, fees stopped being “cost-based” and became “profit-driven.”89
88. See FRB, Profitability of Credit Card Operations of Depository Institutions, 5 (2009), available
at <http://www.federalreserve.gov/Pubs/reports_other.htm> (“Prior to the early 1990s, card
issuers competed primarily by waiving annual fees and providing credit card program enhance-
ments such as airline mileage programs.” Since the early 1990s “interest-rate competition has
played a much more prominent role....Many issuers have attempted to gain or maintain market
share by offering very low, temporary rates on balances rolled over from competing firms or to
select current customers and by offering a wide variety of enhancements.”); Furletti, above
note 16, at 3, 7 (“Consumer awareness of annual percentage rate as a key cost measure, com-
bined with the ability to easily find new card offers and switch issuers, inevitably affected price
competition and rate stickiness.”; “Ultimately, a card’s nominal APR became a competitive
focal point and drove widespread adoption of risk-based pricing.”).
89. On the rise of fees—see Furletti, above note 16, at 11; CardFlash, “Fee Factor 08,” January 14,
2009. See also Furletti, above note 16, at 33, Figure 7 (showing how the “average late fee being
assessed” has increased since 1994). On the move from cost-based to profit-driven fees—see
CardFlash, “Fees and Recession,” December 19, 2008.
Philadelphia Federal Reserve Bank economist, Mark Furletti, who exam-

ined credit card contracts in the late 1990s and early 2000s, concluded that
“risk-based fees have become an important source of revenue for card issu-
ers and have replaced a significant portion of the revenues lost from the
elimination of annual fees and lowered APRs.” And the same is true for
convenience and service fees. Furletti found that these fees were priced “to
provide attractive profit margins.” By 2008, total revenues from cardholder
fees reached $28.9 billion.90
D. Recent Events: The CARD Act

The Credit Card Accountability, Responsibility and Disclosure Act of 2009
affected a major change in the regulation of the credit card market. The
behavioral-economics theory offers predictions about market responses to
the new rules.
The CARD Act imposed restrictions on late and over-limit fees and on
interest rate hikes. This led to a reduction in back-end profits. In response,
issuers would be expected to raise front-end fees and other salient prices,
such as annual fees and purchase APRs. Issuers would also be expected to
cut back on front-end, salient benefits, such as rewards programs, simply
because the back-end funding for these front-end perks is drying up. In
90. On risk-based fees—see Furletti, above note 16, at 11–12. See also Furletti, above note 16, at
10–11 (“With average interest rates on the decline and annual fees becoming unpopular
among their customers, issuers developed more targeted fee structures to replace lost reve-
nues.”); Frontline, “The Credit Card Game,” November 24, 2009 (interview with Shailesh
Mehta, former CEO of Providian: Competition eliminated annual fees, and issuers compen-
sated with penalty pricing). On convenience and service fees—see Furletti, above note 16, at
12. The total revenue from cardholder fees—$28.9 billion—was calculated by subtracting
revenue from interchange fees: $39.1 billion (CardFlash, “Merchant Fees,” January 13, 2009)
from total fee income: $68 billion (CardFlash, “Fee Factor 08,” January 14, 2009). See also
Furletti, above note 16, at 32, Figure 6 (showing a sharp increase in “fee income as a percent-
age of total revenue” since 1996); Improving Federal Consumer Protection in Financial Serv-
ices: Hearing Before the H. Comm. on Financial Services, 110th Cong. app. 94–95 (2007)
(statement of Sheila C. Bair, Chairman, Federal Deposit Insurance Corporation) (noting that
net non-interest income for insured institutions has been growing faster than total net operat-
ing revenue); CardFlash, “Fee Factor 08,” January 14, 2009 (Fee income increased from 17.3
percent of total revenues in 1994 to 40 percent of total revenues, or $68 billion, in 2008.While
these figures include both merchant and cardholder fees, the increase in fee income has been
driven by the rise in late payment fees, over-limit fees, and cash advance fees, coupled with
newer fees such as balance transfer fees, foreign currency exchange surcharges, and expedited
payment fees.) The shift to fees was made possible by the Supreme Court’s 1996 “Smiley”
decision which exempted national card issuers from state restrictions on card fees. CardFlash,
“Fee Factor 08,” January 14, 2009.
c re di t card s 97
addition, issuers would be expected to respond by finding alternative back-

end revenue sources—either by increasing back-end, non-salient fees that
are not covered by the CARD Act or by introducing new, or redesigned,
non-salient terms.91
Existing evidence on the effects of the CARD Act is inconclusive. It may
be too soon to empirically assess the Act’s effect on credit card pricing.
Nevertheless, it is worth noting that industry sources suggest that issuers are
expected to respond—or have already responded—to the Act by increasing
annual fees and purchase APRs. And there is some evidence that issuers are
searching for alternative back-end revenue sources by expanding the use of
non-salient fees and terms not covered by the CARD Act and by introduc-
ing new, non-salient fees and terms.92
V. Welfare Implications
What are the costs of the identified contractual designs, especially when
understood as a response to cardholders’ imperfect rationality? First, com-
plex, multidimensional contracts hinder competition in the credit card mar-
ket. Second, complex and deferred-cost contracts distort the remaining,
weakened forces of competition, leading to excessively high prices on less
salient price dimensions and excessively low prices on more salient price
dimensions. Third, these contractual design features increase the cost of
91. It should be noted that some of these predicted responses are also consistent with a rational-
choice theory of credit card pricing—see Oren Bar-Gill and Ryan Bubb, “Credit Card Pric-
ing: The CARD Act and Beyond” forthcoming in Cornell L. Rev.
92. On the limits of existing evidence on the effects of the CARD Act—see Bar-Gill and Bubb,
id. For industry sources suggesting increases in annual fees and purchase APRs—see Andrew
Martin, “Credit Card Industry Aims to Profit from Sterling Payers,” New York Times, May 19,
2009 (“Banks are expected to look at reviving annual fees…, according to bank officials and
trade groups”); CardFlash, “BofA Changes,” August 18, 2009 (Bank of America expects an
increase in the average APR and a decrease in the variance of APRs); CardFlash,“BofA Basic,”
September 17, 2009 (“The impact of the new credit card rules became clear today as Bank of
America announced plans to launch a new credit card next month with a core interest rate
of prime + 14 percent, compared to the prime + 9.9 percent APR that dominated the market
for the past decade”); CardFlash, “CARD Act Impact,” November 10, 2009 (according to the
Fed’s “Senior Loan Officer Opinion Survey on Bank Lending Practices” from October 2009,
about 40 percent of banks expected to raise annual fees for prime borrowers and about 45
percent of banks expected to raise annual fees for nonprime borrowers). For evidence that
issuers are looking for alternative back-end revenue sources—see Joshua M. Frank, “Dodging
Reform: As Some Credit Card Abuses Are Outlawed, New Ones Proliferate,” Center for
Responsible Lending, 2009, <http://www.responsiblelending.org/credit-cards/research
-analysis/CRL-Dodging-Reform-Report-12-10-09.pdf>.
financial distress. Fourth, the identified contractual designs raise distribu-

tional concerns, as they impose disproportionate burdens on weaker card-
holders. Let’s take a closer look at each one of these costs.
Perhaps the largest cost associated with excessively complex contracts comes
from the inhibited competition that they foster. As described previously,
complexity prevents the effective comparison-shopping that is necessary for
vigorous competition. The market power gained by issuers clearly helps
issuers at the expense of cardholders. But the limited competition also
imposes a welfare cost in the form of inefficient allocation: Cardholders are
not matched with the most efficient issuers.
Complexity harms consumers in yet another way. The imperfectly
rational consumer, and even the rational consumer who declines to invest
the necessary time and money needed to fully understand these complex
contracts, may well end up with a product that is not best for them—
another instance of allocative inefficiency.
Competition in the credit card market focuses on salient dimensions of the
credit card contract. There is much less competition on the non-salient
dimensions. The result is salient provisions that are favorable to cardholders
and non-salient provisions that are unfavorable to cardholders. Is this outcome
necessarily bad for cardholders? The answer is yes. The up-front benefits do
not fully compensate for the back-end costs. Distorted competition in the
credit card market leads to inefficient contracts. Inefficient contracts reduce
both the total surplus created by the issuer–cardholder relationship and the
cardholder’s share of this surplus. (See the example in Section IV.B.1.)
Contracts with up-front benefits and back-end costs are inefficient
because they provide incentives to use the credit card in welfare-
reducing ways. The different dimensions—specifically, the different price
dimensions—of the credit card contract affect a cardholder’s use of the
product. For example, a lower introductory interest rate leads to more short-
term borrowing; a high currency-conversion fee leads to less use of the
credit card outside the United States. A welfare-increasing contract sets
prices efficiently; in other words, to provide incentives for optimal use of
c re di t card s 99
the credit card across the different dimensions. An inefficient contract, in

which prices are set according to their salience instead of the underlying
cost structure, provides the wrong incentives. Using the preceding example,
if short-term credit entails a positive cost to the issuer, but the short-term
interest rate is set at zero, because this dimension is salient to cardholders,
then cardholders will engage in excessive short-term borrowing. If the cur-
rency-conversion fee is set above the cost to the issuer of processing foreign
transactions, because this dimension is non-salient, then cardholders will
engage in suboptimal use of their cards while abroad.
More fundamentally, distorted competition, with the resulting distorted
contracts, can be expected to artificially inflate the demand for credit cards.
Recall that the reason issuers reduce salient prices and increase non-salient
prices is to create the appearance of a cheaper product without actually
offering a cheaper product. When misperception leads cardholders to
underestimate the total cost of the credit card (or overestimate the net
benefit from the card), then demand for credit cards will be too high. Add
to that the low, even negative, per-transaction fee, and we get too many
credit cards used too often (for transacting).93
C. Financial Distress
There are two distinct arguments about the relationship between credit
cards and financial distress.
The first argument is that credit cards cause financial distress. There is
empirical evidence from three independent studies conducted by Ronald
Mann, Michelle White, and Atif Mian and Amir Sufi, suggesting a causal
link between credit card debt and consumer bankruptcies.94 To the extent
that credit cards and inefficient credit card contracts increase the risk of
financial distress, they impose a cost on cardholders and on society.
The second argument is that, while not causing financial distress, credit
cards, with their distorted contractual design, exacerbate the cost, to card-
holders, from financial distress. Deferred-cost contracts often defer the costs
to periods when the cardholder is in financial distress. The financially
distressed cardholder is more likely to incur a late payment fee, pay an
93. See Sujit Chakravorti and William R. Emmons, “Who Pays for Credit Cards?” (2001) Fed.
Reserve Bank of Chi. Emerging Payments Occasional Paper Series, No. 1, at 1(it may well be that
credit cards are “overused”).
94. See above Section I.C.
over-limit fee, and trigger a default interest rate. Economic theory suggests
that paying these high prices when in financial distress is especially painful,
because of the decreasing marginal utility from money.
The abstract principle of decreasing marginal utility from money reflects
the very concrete real-world experience of credit card holders. When a
consumer is employed in a well-paid job, saving money on annual fees, per-
transaction fees, and short-term interest rates is a nice but insignificant perk.
But when the consumer is between jobs, struggling to make ends meet, or
facing the financial burden of unexpected and mounting medical bills, pay-
ing this same amount of money (realistically, much more) in penalty fees
and rates would likely be quite painful.
D. Distributional Concerns
Typically, consumers who benefit from rewards programs and frequent-flyer
miles are not the same consumers who pay penalty fees and default interest
rates. As we have seen, long-term prices finance short-term perks. We have
not yet addressed the question of whether the same group of consumers pay
the long-term prices and enjoy the short-term perks. There is reason to
believe that there is only a partial overlap between the paying group and the
benefiting group. There is also reason to believe that socio-economic status
is not randomly divided across the two groups. The distorted pricing in the
credit card market leads to regressive cross-subsidization between the differ-
ent consumer groups.95
The cross-subsidization argument assumes that short-term perks for the
rich are funded by the high, long-term prices that the poor pay. This argu-
ment needs to be qualified. The analysis in this chapter focuses on the
issuer–cardholder relationship, sidestepping the important role of the mer-
chant who accepts the card and transfers a percentage of the transaction
amount (the interchange fee) to the issuer. It is possible that short-term
perks for the rich are funded by the interchange revenues from the high-
volume transactions of these rich cardholders. But this raises a different
95. See Federal Reserve Bank of Boston, “Consumer Behavior and Payment Choice: 2006 Con-
ference Summary” (2007) Public Policy Discussion Paper 07–4, 49, available at <http://
www.bos.frb.org/economic/ppdp/2007/ppdp0704.htm> (hereinafter “FRB Boston Con-
ference Proceedings”) (summarizing the proceedings of the second Consumer Behavior and
Payment Choice conference, held at the Federal Reserve Bank of Boston on July 25–27, 2006,
where participants noted “that non-revolving credit card users who have rewards programs
are subsidized by revolving users without rewards programs”).
c re di t card s 101
distributional concern: If merchants increase retail prices to compensate for

the interchange tax that they transfer to issuers, then all customers—includ-
ing poorer cash customers—end up funding those short-term perks for the
rich.96 In any event, the weaker cardholders will still pay more of the high
long-term prices and enjoy less of the short-term perks, even if the short-
term perks are not funded by revenues from the long-term prices.
VI. Market Solutions

The welfare costs of credit cards designed in response to consumer biases
can be large. Legal policy intervention may be needed to curb these welfare
costs. But before considering legal intervention, we should explore the abil-
ity of the market to overcome the behavioral market failure on its own.
Such market solutions exist. Some issuers offer credit card products that do
not exhibit the distorted contract design described above. Perhaps most
importantly, the rise of debit cards can be seen as a market solution to the
problems with credit cards. Many of the misperceptions that are prevalent in
the credit card market are borrowing-related. By removing the borrowing
service, debit cards avoid triggering these misperceptions.97
Ultimately, however, these market solutions are imperfect because they
cater to only a limited slice of the credit card market; namely, the more
sophisticated consumers.The underlying problem is that markets respond to
consumer demand. If imperfectly rational consumers demand distorted
products, issuers will meet this demand. The solution, then, must focus on
the demand problem. Consumers can (and often do) overcome their mis-
takes and biases, but learning is slow and not everyone learns.98 Also, issuers
have insufficient incentives to educate consumers and help them overcome
their imperfect rationality. Consequently, less sophisticated consumers will
continue to purchase the distorted products. Only the more sophisticated,
96. It is not clear that merchants will in fact increase retail prices to compensate for the inter-
change tax. See, e.g., Steven Semeraro, “The Reverse-Robin-Hood-Cross-Subsidy Hypoth-
esis: Do Credit Card Systems Effectively Tax the Poor and Reward the Rich?”, Rutgers L.J.,
40 (2009), 419.
97. Prepaid cards should also be mentioned as a potential market solution.
98. See Sumit Agarwal, John C. Driscoll, Xavier Gabaix, and David I. Laibson, “Learning in the
Credit Card Market”, February 8, 2008, available at SSRN: <http://ssrn.com/abstract=1091623>
(Borrowers learn to avoid fees, but learning is imperfect and borrowers gradually forget what
they have learned.)
rational consumers (and those who are aware of their imperfect rationality)
will be attracted to the non-distorted products, to the market solutions.99
A. Consumer-Friendly Credit Cards

The credit card market is a large, diverse market. The preceding analysis
focused on design features of the common credit card contract. But not all
cards are created equal, nor do all cards exhibit all of the problematic design
features highlighted in previous sections. Some issuers, in fact, offer
consumer-friendly cards.
There are, for example, credit cards with a less complex, more transparent
contract design than has been described previously. For example, in 2009,
Bank of America introduced its “BankAmericard Basic,” which has one rate
for all transactions and no over-limit fees. In 2011, Citi introduced the “Sim-
plicity Card,” featuring a single rate across all purchases, balance transfers,
and cash advances—with no late fees or penalty rate.100
Not all cards defer costs to the same extent. A recent study by Ryan
Bubb and Alex Kaufman found that credit cards offered by credit unions
exhibit less cost deferral than credit cards offered by investor-owned issu-
ers. Bubb and Kaufman explain that credit unions have less incentive to
make a profit at the expense of their members. The credit union’s owner-
ship structure serves as a commitment to spare the consumer the high
prices affixed to non-salient dimensions. Bubb and Kaufman argue that
more sophisticated consumers, who are aware of their biases, are attracted
to credit unions. While clearly an optimistic finding, this optimism is tem-
pered by the limited market share of credit unions, which stands at less
than 5 percent.101
99. This problem restricts the impact of another market solution—third-party intermediaries,
like creditcards.com and creditkarma.com, that provide easy comparison among card offers.
Efficient comparisons will not help, if the consumer is comparing the wrong product
dimensions or not placing proper weights on the different product dimensions. See section
V.B. (Distorted Competition).
100. On Bank of America offerings—see CardFlash, “BofA Clarity,” November 30, 2009. Bank of
America also made a “Credit Card Clarity Commitment,” as part of its campaign to simplify
loan information for consumers, and mailed out a one-page summary of customers’ rates,
fees, and payment information to 40 million current cardholders. Id. On Citi’s offerings—see
CardFlash, “Citi Offers Credit with Single APR and No Late Fees,” July 26, 2011.
101. See Ryan Bubb and Alex Kaufman, “Consumer Biases and Firm Ownership,” (2009) Work-
ing Paper. The market share of credit unions—less than 5 percent—is based on 2006 data
from the 2007 Card Industry Directory (by SourceMedia, Inc.).
Good behavior among credit card issuers is not limited to credit unions.
For example, before the practice was banned by the CARD Act, Chase
voluntarily eliminated double-cycle billing. Several issuers offer online
financial education and specialized tools to help consumers manage their
credit. For example, in 2010, Discover Financial Services launched its
“Straight Talk” consumer website, designed to help consumers more easily
understand how credit cards work. Discover also offers a variety of tools and
resources to help consumers manage their credit, including “The Spend
Analyzer,” “The Paydown Planner,” and “The Purchase Planner.” Other
issuers offer similar tools. Some also offer an automatic payment service,
which can eliminate the risk of late payment.102
B. Debit Cards
Perhaps the solution to the problems inflicting the credit card market lies in
a sister product: the debit card. Debit cards have enjoyed substantial growth
in recent years.This growth has come, at least in part, at the expense of credit
cards.103 As noted earlier, debit cards are used only for transacting and thus
offer a more consumer-friendly product when biases are borrowing-related.
To understand the rise of the debit card and to assess whether debit cards
can reduce the welfare costs in the credit card market, we need to under-
stand the drivers of the demand for debit cards. At first blush, the growth in
debit card use is puzzling, at least to the extent that this growth is fueled by
consumers who have, or can get, a credit card. After all, the credit card is a
better product. It does everything that a debit card does and more. A con-
sumer who wants a convenient, secure payment method can use a credit
card and pay her balance in full each month. The credit card allows you to
102. See CardFlash, “Chase Pricing,” November 20, 2007; CardFlash, “Straight Talk,” February 23,
2010; CapitalOne, Credit Card Payment Calculator, <http://www.capitalone.com/
calculator/>;American Express,A Notice AboutYour Account Alerts, email message received
by Oren Bar-Gill, July 29, 2009 (introducing the Spend Tracking Alert and Balance Tracking
Alert).
103. See CardFlash, “Visa Debit,” May 5, 2009. (In 2002, U.S. Visa debit transactions surpassed
credit transactions for the first time. In the fourth quarter of 2008, 70 percent of U.S. Visa
transactions were debit transactions. Also, the fourth quarter of 2008 was the first period
when debit cards surpassed credit cards in U.S. Visa payment volume: Debit payments vol-
ume was $206 billion, compared with credit payments volume of $203 billion.) A large part
of debit card growth came at the expense of cash and checks. See FRB Boston Conference
Proceedings, above note 95. But debit cards are also replacing credit cards. See Jonathan
Zinman, “Debit or Credit?” (2006) Dartmouth College Economics Department Working
Paper 3–4.
borrow; it doesn’t force you to do so. In fact, credit cards are a better choice
than debit cards for consumers who wish to transact but not to borrow, as
they offer superior fraud-protection.104
Credit cards are also better than debit cards for non-borrowers, as they
provide more lucrative rewards programs and other perks, like extended
warranties and car rental insurance.105 The behavioral-economics theory
provides one explanation for the added benefits that credit cards offer to
non-borrowers: The back-end revenues that credit card issuers obtain from
borrowers on the non-salient, borrowing-related terms help finance the
greater front-end benefits and perks offered to non-borrowers.These advan-
tages of credit cards may explain the substantial delay in the introduction of
debit cards in the United States.106
Still, the number and transacting volume of debit cards are growing fast.
Why? One explanation is that consumers are becoming more sophisticated
and less optimistic about their willpower, their ability to avoid financial
hardship, and the probability of inadvertently triggering hidden fees. These
consumers, armed with their newfound awareness, want to make a “pre-
commitment” to stay away from the seduction of credit card borrowing and
104. See Mann, above note 22 (debit cards offer fewer legal protections); Privacy Rights Clear-
inghouse/UCAN, “Paper or Plastic? What Have You Got to Lose” March 2011, available at
<http://www.privacyrights.org/fs/fs32-paperplastic.htm>.
105. Rewards on debit cards, while growing, are not as prevalent or as generous as credit card
rewards. See FRB Boston Conference Proceedings, above note 95, at 22–3 (Debit card
reward programs, while still relatively rare, are increasing rapidly); Andrew Ching, and
Fumiko, Hayashi, “Payment Card Rewards Programs and Consumer Payment Choice” (May
13, 2008), available at SSRN: <http://ssrn.com/abstract=1114247>; CardFlash, “Debit
Cards,” August 18, 2008 (the “2008 Debit Issuer Study,” conducted by Oliver Wyman for
PULSE and covering more than 74 million debit cards issued by 62 financial institutions—
including large banks, community banks, and credit unions—revealed that 51 percent of
respondents now offer cards with rewards programs, compared to 37 percent in 2006); Card-
Flash, “Debit Rewards,” August 28, 2009 (based on First Data’s “Customer Loyalty Study”—
in 2009, membership in debit card reward programs increased to 45 percent compared to
34 percent in 2008, but only 23 percent of respondents indicated that the benefits available
through the reward program was “very influential” in their choice of financial institution).
106. See Evans and Schmalensee, above note 5, at 55, 76–7 and ch. 12 (“[D]ebit cards languished
[in the United States] until the mid-1990s.”). Debit cards have their own back-end costs,
specifically overdraft fees. See CardFlash, Debit Overdraft, July 12, 2007 (a Center for
Responsible Lending study found that debit card overdrafts are now the single largest source
of overdraft fees. The study also found that consumers paid about $17.5 billion in fees for
$15.8 billion in abusive overdraft loans.) Revenues from these overdraft fees may have helped
fund the debit card reward programs, at least until they were limited by regulation that
requires banks to obtain explicit consent before enrolling a consumer in an overdraft protec-
tion plan. See FRB, “What You Need to Know: New Overdraft Rules for Debit and ATM
Cards, <http://www.federalreserve.gov/consumerinfo/wyntk_overdraft.htm>.
the costs it entails. Sophisticated consumers wish to tie their own hands.
Debit cards provide the rope.
There are two problems, however, with debit cards as a market solution. The
first is that the debit card solution is limited in scope. Only sophisticated con-
sumers who understand the risks of credit cards will choose debit cards instead.
Using back-end revenue to finance lucrative front-end perks, credit card issu-
ers will be able to prevent the less sophisticated consumers from switching to
debit cards. The second problem is that the shift to debit cards, to the extent
that it occurs, is a second-best solution. Debit cards can be better than bad
credit cards—those designed in response to consumer misperception. But they
are not necessarily better than good credit cards. Credit cards can provide low-
cost, convenient access to credit, which can be beneficial to many consumers.
The goal should not be to abolish credit cards, but to reshape the credit
card product. Policymakers can play an important role in pushing the mar-
ket in the right direction.
VII. Policy Implications: Rethinking Disclosure
A. Focusing on Disclosure
The welfare costs detailed above and the limits of market solutions provide
the basis for considering legal intervention in the credit card market. The
proposed behavioral-economics theory provides guidance for the design of
potential policy responses. Here, I focus on one regulatory technique;
disclosure mandates.
Any discussion of credit card regulation must note the important reforms
put in place by the CARD Act of 2009. The CARD Act, and its imple-
menting regulations, enhanced the mandatory disclosure regime governing
credit cards, as we will see shortly. But unlike most previous legal interven-
tions in the credit card market, the CARD Act did not stop at disclosure. It
imposed substantive restrictions, banning practices and limiting prices. Some
of these restrictions are clearly welfare-enhancing, such as the prohibition
on allocating payments to low-interest balances first. Others, like the restric-
tions on risk-based pricing, are more controversial.These debates are beyond
the scope of this Part, which, as noted earlier, focuses on disclosure.
Any treatment of disclosure as a regulatory technique should begin with
the question of efficacy: Are disclosure mandates able to affect behavior and
improve market outcomes? In the credit card market, evidence suggests that
disclosure has been effective, at least to some extent. A series of studies by
Federal Reserve economists have found that Truth-in-Lending disclosures
have raised consumer awareness of the APR as a key determinant of the cost
of credit. Moreover, these studies found that the increased salience of the
APR has led to competition on this price dimension.107
There is also evidence that the enhanced disclosures mandated by the
CARD Act have been effective. Specifically, a Consumer Reports survey
found that 23 percent of respondents were motivated by the new Minimum
Payment Warning on their bills to pay off their credit cards faster.108 But
even if the efficacy of current disclosures is limited, this still would not con-
demn disclosure as a regulatory strategy, because existing disclosure man-
dates are not optimally designed. In this Part, we’ll look at ways regulators
can design better disclosure mandates for the credit card market.
B. Disclosing Product-Use Information

The preceding analysis suggests that consumer misperception is responsible
for the distorted design of the credit card contract. The problem, in essence,
is that when shopping for a credit card, consumers are not armed with a
sufficiently accurate assessment of how they will use the credit card product.
Legal policy can address this problem. If consumers misperceive their use
patterns, legislators and regulators should require issuers to provide prod-
uct-use information to consumers.
While disclosure mandates are perhaps the main regulatory technique
used in the credit card market, existing mandates focus almost exclusively
on the disclosure of product-attribute information. Consider the Truth-in-
Lending Act (TILA) and Regulation Z, which require specific disclosures in
credit card applications and solicitations. These rules mandate disclosure of
product-attribute information, specifically interest rates and fees.109
Disclosure regulation should go beyond the current scope of TILA.
The goal is not only to educate consumers about credit terms, but also to
107. See Furletti, above note 16, at 9; Thomas A. Durkin, “Credit Cards: Use and Consumer Atti-
tudes, 1970–2,” Fed. Reserve Bull., September 2, 623–34; Jinkook Lee and Jeanne M. Hogarth,
“The Price of Money: Consumers’ Understanding of APRs and Contract Interest Rates”, J.
of Pub. Pol’y and Marketing, 18 (1999), 66.
108. See CardFlash, “Consumer Reports,” October 20, 2010.
109. 15 U.S.C. § 1637(c); 12 C.F.R. §§ 226.18, 226.5a.
educate them about their own preferences and cognitive biases. Knowledge
of credit terms is meaningless if the consumer mistakenly precludes the pos-
sibility of a future need to borrow. Similarly, knowing the magnitude of a
late fee is not helpful if the consumer mistakenly believes that he or she will
never pay late.110
The declared purpose of TILA is “to assure a meaningful disclosure of
credit terms . . . and to protect the consumer against inaccurate and unfair bill-
ing and credit card practices.” The Interpretive Notes and Decisions accom-
panying TILA’s opening section elaborate that the purpose of TILA is “to
help correct what Congress perceived as widespread consumer confusion
about the nature and cost of credit obligations.”111 Confusion about the cost
of credit can persist, even when consumers know all relevant product attributes.
“[M]eaningful disclosure” should include product-use information.
C. Designing Product-Use Disclosures

Designing disclosures that incorporate product-use information is not a
trivial task. Product-use information can be disclosed at different levels.
Issuers can be required to disclose—
• average-use information, where use patterns are averaged out across the
entire population of consumers or, preferably, across some demographic
or socio-economic subgroup of consumers; or
• individual-use information, based on historic use-pattern information
collected on the individual consumer.
Individual-use information, when available, is preferable to statistical,
average-use information. Existing evidence suggests that disclosure of sta-
tistical evidence is not always convincing. In particular, if the disclosed
average level of borrowing or average number of late payments per year is
taken across a large group of consumers, any individual consumer may
believe that she is among the few who will not borrow and never pay late.
Individual-use information is immune against such “I am better than
110. See Furletti, above note 16, at 5 (“Recent changes in how issuers price credit cards, however,
have resulted in new levels of pricing complexity and created a structure of credit costs that
can impact some customers very differently than others. That is, the cost that a consumer
faces greatly depends on the way he or she uses the credit card.”); CardFlash, “Reward Card
Analysis,” June 2, 2008 (consumers must know their use patterns in order to choose the
optimal rewards card for themselves).
111. See 15 U.S.C. § 1601(a) and the accompanying Interpretive Notes and Decisions.
average” reasoning. When confronted with disclosures about their own

levels of borrowing and the number of late payments that they made, con-
sumers are not able to dismiss the information as an abstract, irrelevant
statistic. Still, even individual-use information may not be entirely con-
vincing, as the consumer could believe that past borrowing is not indica-
tive of future choices.112
Importantly, issuers have a lot of individual-use information, which they
collect and use to maximize profits. Duncan McDonald, former general
counsel of Citigroup’s Europe and North America card businesses, noted:
“No other industry in the world knows consumers and their transaction
behavior better than the bank card industry. It has turned the analysis of con-
sumers into a science rivaling the studies of DNA. The mathematics of virtu-
ally everything consumers do is stored, updated, categorized, churned, scored,
tested, valued, and compared from every possible angle in hundreds of the
most powerful computers and by among the most creative minds anywhere.
In the past 10 years alone, the transactions of 200 million Americans have been
reviewed in trillions of different ways to minimize bank card risks.”113
Issuers should be required to share this information with their customers.

Optimal disclosure design should also consider the advantages of com-
bining product-use information with product-attribute information. For
example, providing information on the number of late payments made dur-
ing the past year is helpful. Providing information on the total amount
of money paid in late fees during the past year is even more helpful. This
112. On the limits of average-use disclosures—see Christine Jolls et al., “A Behavioral Approach
to Law and Economics”, Stan. L. Rev., 50 (1998), 1471, 1542 (people tend to underestimate
their future risks even if they actually understand average risks); and compare: Svenson (fn
15, Chapter 1, this volume) 117 (a vast majority of drivers believe that their driving skills are
above average). On the advantages of individualized disclosure—see generally Cass Sunstein
and Richard Thaler, Nudge (describing their RECAP proposal) (fn 31, Chapter 1, this vol-
ume). See also Barry Nalebuff and Ian Ayres, Why Not? (Harvard Business School Press,
2003) 181 (arguing that issuers should be required to disclose to consumers the likelihood
that they will incur late and over-limit fees, preferably based on individual data that the issuer
collects on the specific consumer).
113. Duncan MacDonald, “Viewpoint: Card Industry Questions Congress Needs to Ask,”
American Banker, March 23, 2007. See also FRB, “Report to the Congress on Practices of
the Consumer Credit Industry in Soliciting and Extending Credit and their Effects on
Consumer Debt and Insolvency,” 19 ( June, 2006), available at <http://www.federalreserve
.gov/boarddocs/rptcongress/bankruptcy/bankruptcybillstudy200606.pdf> (detailing the
huge amounts of information that issuers have and use); Charles Duhigg, “What Does
Your Credit-Card Company Know about You?” New York Times, May 17, 2009 (describing
the vast amount of information, especially product-use information, that credit card com-
panies collect, and then analyze using sophisticated algorithms informed by psychology
research).
disclosure combines product-use information (the number of late payments)

with product-attribute information (the magnitude of the late fee).
Timing is another important dimension in designing disclosure man-
dates, specifically when use information is concerned. When should the
information be disclosed? Perhaps the easiest way to provide use informa-
tion is on the monthly statement. As the consumer uses the card, the issuer
collects and discloses use information. Recent reforms have substantially
improved disclosures on the monthly statement, including disclosures that
combine attribute and use information. Under FRB regulations, which
took effect along with the CARD Act implementing rules, issuers must
disclose, on the monthly statement, both monthly and year-to-date totals of
interest charges and fees, separately.114
But disclosures on the monthly statement are not enough. To ensure
competition and positive market outcomes, the consumer should be able to
compare a current credit card to others being offered. Disclosures on the
monthly statement, when it incorporates use information, tells consumers
how much they are paying on their current card. In addition, consumers
need to know how much they would pay if they decide to switch to a dif-
ferent card. This takes us from monthly statement disclosures to disclosures
in advertisements and solicitations. Under existing law, these disclosures do
not contain any use information, the idea being that the new issuer does not
have any use information to disclose.115
Product-use information can be included in advertisements and solicita-
tions. One option is to provide statistical use information or combine statis-
tical use information with attribute information.116 Another, more promising,
option is to utilize individual-use information from the consumer’s previ-
ous experiences, if any, in the credit card market. Under this option, the
current issuer would be required to provide its customers with an electronic
file containing all the use information that it collected. This might take the
form of an Excel spreadsheet, for instance, with the consumer’s individual-
114. See CFPB, CARD Act Factsheet, above note 4.

115. See Elizabeth Renuart and Diane E.Thompson, “The Truth,The Whole Truth, and Nothing
but the Truth: Fulfilling the Promise of Truth in Lending”, Yale J. Reg., 25 (2008), 181, 188
(“Fees are presumed to be unknown”).
116. Compare: Renuart and Thompson, id., at 189 (describing a proposal by the National Con-
sumer Law Center to require disclosure of a “typical” APR, calculated as an average of
effective APRs paid by borrowers—these effective APRs include fees, such as late fees and
over-limit fees, that are not included in the APR disclosed on advertisements, solicitations,
and at account opening).
use data from the past three years. Consumers could take this file to the new
issuer and get a total price quote, based on their use patterns. Such elec-
tronic disclosure can also facilitate the work of intermediaries. Companies
like BillShrink can combine the use information with attribute information
and recommend the card that best fits the consumer’s needs.
Disclosures in advertisements and solicitations and on the monthly state-
ment provide two options for the timing of disclosure. A third option, real-
time disclosure, is particularly promising with respect to product-use
information. In certain cases, disclosure can be more effective if the infor-
mation is provided before the end of the billing cycle. According to a pro-
posal by Ronald Mann, issuers would be required to provide point-of-sale
information. For instance, if by charging a current purchase to a certain
credit card the consumer would exceed the credit limit on that card and
incur a fee, this information would be provided to the consumer, allowing
him or her to use a different card or an alternative payment system.117
Finally, optimal disclosure design must address concerns about informa-
tion overload. This is a general concern that applies to both product-attribute
and product-use disclosures. There are two responses to this concern. One is
to design simple disclosures. The year-to-date total interest paid is an exam-
ple of such a simple disclosure. The idea is to provide highly aggregated
information that can be understood and applied by the imperfectly rational
consumer. The second approach is to provide much more information—raw,
disaggregated information—but not to consumers. Recall the Excel file with
detailed use information. The consumer wouldn’t be expected to read and
analyze the information in that file. Rather he or she would take the file to
a competing issuer or to an intermediary for processing.
D. Steps in the Right Direction

Legislators and regulators are starting to understand the importance of
product-use disclosures. The Dodd–Frank Wall Street Reform and Con-
sumer Protection Act of 2010 imposes a general duty, subject to rules pre-
scribed by the new Consumer Financial Protection Bureau, to disclose
information, including usage data, in markets for consumer financial
products.118
117. See Mann, above note 22.

118. Pub. L. 111–203, Title X, Sec. 1033.
Focusing on more concrete steps, the CARD Act requires that issuers
disclose a Minimum Payment Warning on the monthly bill that includes
information on the amount of time it will take to pay off the balance and
the aggregate total payment if only the minimum amount is paid each
month. The CARD Act also requires that issuers calculate and disclose the
monthly payment that would pay off the cardholder’s balance in three years,
as well as the savings—in total payments—from this faster repayment
schedule.119 The minimum payment disclosures address the slow repayment
problem identified earlier. They combine product-attribute information
with certain use patterns specified by the Act and implementing regula-
tions—comparing slow repayment (making only the minimum payment)
to faster repayment (paying off the balance in three years).
Recent regulations also mandate that issuers disclose, on the monthly
statement, monthly and year-to-date totals of interest charges and fees,
separately.120 As mentioned above, this is an example of a disclosure that
combines individual-use information and product-attribute information.
While promising, this disclosure can and should be improved. First, disclos-
ing a single total cost figure can be more effective than disclosing two sep-
arate figures—one for interest and one for fees. Second, year-to-date figures
make sense for monthly statement disclosures but contain limited use infor-
mation. Issuers could be required to provide a year-end summary with total
annual cost figures that are based on a longer history of use patterns—three
years, perhaps. Finally, to facilitate competition, use-based information needs
to be disclosed also by new issuers. For this purpose, regulators should
require existing issuers to provide, in electronic form, detailed use informa-
tion that could be then transferred to new issuers or to intermediaries.
Disclosing use-based, total-cost information, by both existing and new
issuers, can potentially eliminate the complexity and cost-deferral problems.
Consumers should not care about complexity if issuers are required to dis-
close the bottom-line, aggregate costs. Moreover, issuers will have no incen-
tive to artificially increase complexity, only to disguise the true cost of credit,
119. See FRB, “Credit Card Rules, February 22”, above note 34 (describing the new CARD Act
rules).
120. See CFPB, CARD Act Factsheet, above note 4. It is important that all fees be included.
Otherwise, issuers will have an incentive to increase excluded fees. Cf. Renuart and
Thompson, above note 115, at 185, 203 (explaining the adverse effects of an APR disclosure
that excludes certain fees, Renuart and Thompson note: “The excluded fees can contribute
mightily, yet invisibly, to the cost of credit. As a result, lenders are motivated to structure their
credit transactions to take advantage of these exclusions.”).
since they must—by law—disclose the total cost of credit anyway.121 A total
cost disclosure can also address the deferred-costs problem. The total cost
measure would take into consideration the temporal dimension of credit
card pricing, combining short-term and long-term costs. Accordingly, con-
sumers who shop for credit based on the total-cost figure will not underes-
timate long-term costs. And, as a result, issuers will have no incentive to
defer costs.
Conclusion
Credit card contracts are complex and multidimensional. Credit card pric-
ing is, in many cases, salience-based rather than cost-based. This contract
design is the product of a behavioral market failure, a result of the interac-
tion between consumer psychology and market forces. The market failure
stifles competition by increasing the costs of comparison shopping. It dis-
torts competition because, to attract imperfectly rational cardholders, issuers
compete by reducing the perceived total price rather than the actual total
price. The costs of financial distress are increased, and distributional con-
cerns are raised.
The identification of a market failure—the behavioral market failure—
opens the door for legal intervention. Recent reforms, specifically the
CARD Act and the Dodd–Frank Act, take important steps in the right
direction. Focusing on the design of optimal disclosure mandates, this chap-
ter provided guidance to lawmakers, including those entrusted with the
implementation of the CARD Act and the Dodd–Frank Act, on how to
most effectively address the behavioral market failure.
121. Board of Governors of the Federal Reserve System and the Department of Housing and
Urban Development, “Joint Report to the Congress Concerning Reform to the Truth and
Lending Act and the Real Estate Settlement Procedures Act” 9 (1998), available at <http://
www.federalreserve.gov/boarddocs/press/general/1998/19980717/default.htm> (discussing
the benefits of the APR as a total cost of credit measure, the Report notes: “The APR con-
cept deters hidden or ‘junk’ fees to the extent that the fees must be included in the APR
calculation.”).
Appendix: Teaser Rates

Why are teaser rates so effective? The answer is that a consumer with a current
financing need will take the teaser-rate “bait.” For such consumers, who have
already decided to incur debt on the new credit card, the interest rate on this debt
will be important. Moreover, if the new card permits balance transfers from old
cards, the switch may present a current benefit. On the other hand, if consumers
incorrectly believe that they will not borrow beyond the introductory period, they
will not mind the steep jump in the interest rate from the low teaser rate to the high
post-introductory level.122
To better understand the operation of teaser rates, let us return to the
hyperbolic discounting model. Consider two pairs of points on a timeline
starting at T = 0, such that the temporal distance between the two points in the
first pair is equal to the temporal distance between the two points in the second
pair, and the first pair is closer to T = 0 relative to the second pair. Hyperbolic
discounting implies that the discounting, from the T = 0 perspective, between
the two points in the first pair will be greater than the discounting between the
two points in the second pair. Consequently, the likelihood of a preference
reversal is greater for the temporally distant pair. Returning to teaser rates, this
implies that the likelihood of unanticipated borrowing is increasing in the
temporal distance between T = 0 and the point in time when the actual
borrowing decision will be made. This means that consumers are less likely to
underestimate their short-run level of borrowing, which would make them
more sensitive to short-term interest rates.
Figure 2.2 offers a graphic illustration of short-run borrowing (T = 1,2), as
compared to long-run borrowing (T = 3,4).While a preference reversal is obtained
with respect to long-run borrowing, leading to underestimation of future borrowing,
there is no preference reversal with respect to short-run borrowing. The consumer
believes at T = 0 that he or she will borrow, and indeed borrows at T = 1. Since there
is no underestimation of short-run borrowing, this consumer would be more
sensitive to short-run interest rates. This explains the competition between credit
card issuers on the teaser-rate dimension.
Another related explanation for the prevalence of teaser rates is based on the
concept of switching costs. In its simple form, the argument is that the costs of
switching from one credit card to another prevent such switching, at least to a
certain degree.Therefore, issuers can lure consumers with low introductory interest
rates, counting on switching costs to prevent at least some consumers from switching
122. See Ausubel, above note 10, at 262–3.

Cost of Cost of
borrowing borrowing
Benefit Benefit
from from
borrowing borrowing
T
0 1 2 3 4
Figure 2.2. Borrowing in the near versus the more distant future
to another card once the introductory period is over.123 The problem with this
simple version of the switching-cost argument is that rational consumers anticipate
the lock-in effect. Recognizing that they would not switch to a new card, consumers
would weigh also the high, post-introductory interest rates when choosing among
competing credit card products.
The behavioral-economics theory suggests a more persuasive version of the
switching-cost story. Even if consumers anticipate lock-in, they still underestimate
the cost of lock-in, since they do not expect to borrow (or to borrow as much) in
the future. Hyperbolic discounting reinforces this revised version of the switching-
cost argument. Naive hyperbolic discounters may wrongly anticipate that they
would switch to a new card, but in fact will not switch when the introductory
period ends. From an ex ante perspective, when both the switching costs and the
benefits of switching lie in the future, a naive hyperbolic discounter might believe
that she will switch to a new card at the end of the introductory period offered by
123. See Ausubel, above note 10, at 263 (“[E]conomic theory suggests that firms in a market with
substantial search/switch costs will find advantage in utilizing introductory offers or sign-up
bonuses to lure new customers.”). In the credit card market there may be non-trivial switch-
ing costs. Shui and Ausubel estimated the cost of switching, including psychological costs, at
$150. See Shui and Ausubel, above note 28. See also National Consumer Law Center, Truth
in Lending (4th edn., 1999) 262. Rewards programs based on the accumulation of points or
frequent-flyer miles generate additional switching costs. Also, switching becomes extremely
difficult as the size of a borrower’s outstanding balance increases. See Ronald J. Mann,
“ ‘Contracting’ for Credit”, Mich. L. Rev., 104 (2006), 899, 925; Paul S. Calem et al., “Switch-
ing Costs and Adverse Selection in the Market for Credit Cards: New Evidence” J. Banking
and Fin., 30 (2006), 1653, 1655; Victor Stango, “Pricing with Consumer Switching Costs:
Evidence from the Credit Card Market”, J. Indus. Econ., 50 (2002), 475, 477, 479–80; Ronald
J. Mann, “Bankruptcy Reform and the ‘Sweat Box’ of Credit Card Debt”, U. Ill. L. Rev., [vol.
no] (2007), 375, 389. Interestingly, while the teaser rate strategy surely relies on switching
costs to limit defection at the end of the introductory period, one of the main purposes of
teaser rates is to induce switching.
the old card. However, when the introductory period offered by the old card ends,
the switching costs are imminent while the benefits from switching still lie in the
future. Thus, applying the high short-term discount rate, the consumer may decide
not to switch.This more sophisticated version of the switching-cost argument helps
explain consumers’ heightened sensitivity to teaser rates and the resulting prevalence
of teaser rates in credit card offers.124
124. See DellaVigna and Malmendier, above note 39, at 26–7 (naive agents underestimate the
probability that they will “renew the contract,” i.e. continue to borrow after the introductory
period); Ran Spiegler, Bounded Rationality and Industrial Organization (Oxford University
Press, 2011), ch. 2 (describing teaser rates as a contract design response to demand generated
by naive consumers).
3
Mortgages
Introduction
Almost 3 million subprime loans were originated in 2006, bringing the total
value of outstanding subprime loans to over a trillion dollars.1 A few months
later, the subprime crisis began. Foreclosure rates soared. Hundreds of bil-
lions—perhaps trillions—of dollars were lost by borrowers, lenders, neigh-
borhoods, and cities, not to mention broader effects on the U.S. and world
economies.2
In this chapter, we’ll examine the subprime mortgage contract and its
central design features. As we will see, for many borrowers these contractual
design features were not welfare maximizing. In fact, to the extent that the
design of subprime mortgage contracts contributed to the subprime crisis,
this welfare loss to borrowers—substantial in itself—is compounded by
much broader social costs. A better understanding of the market failure that
1. See Yuliya Demyanyk and Otto Van Hemert, “Understanding the Subprime Mortgage Crisis”
Rev. Fin. Stud., 24 (2011), 1848, 1853, 1854 (tbl. 1) (analyzing data covering approximately 85
percent of securitized subprime loans. In 2006, 75 percent of subprime loans were securitized,
and the authors’ data set included 1,772,000 subprime loans originated in 2006, implying a
total of 1,772,000/(0.85 * 0.75) = 2,779,608); State of the U.S. Economy and Implications for
the Federal Budget: Hearing before the H. Comm. on the Budget, 110th Cong. 10 (2007)
(hereinafter “Hearing”) (prepared statement of Peter Orszag, Director, Congressional Budget
Office) (“By the end of 2006, the outstanding value of subprime mortgages totaled more than
$1 trillion and accounted for about 13 percent of all home mortgages.”); Center for Respon-
sible Lending, “A Snapshot of the Subprime Market” <http://www.responsiblelending.org/
issues/mortgage/quick-references/a-snapshot-of-the-subprime.html> (estimating that as of
November 27, 2007, there were 7.2 million outstanding subprime loans with an estimated total
value of $1.3 trillion).
2. See Cong. Budget Office, The Budget and Economic Outlook: FiscalYears 2008 to 2018 (2008) 23 (here-
inafter “CBO Outlook”), available at <http://www.cbo.gov/ftpdoc.cfm?index=8917andtype=1>
(noting estimates of between $200 billion and $500 billion for total subprime-related losses and
noting the additional—and potentially substantial—indirect adverse effects of the subprime crisis
on the economy).

mortgage s 117
produced these inefficient contracts should inform the ongoing efforts to

reform the regulations governing the subprime market.
A. Contract Design
During the five years preceding the crisis, the subprime market experienced
staggering growth as riskier loans were made to riskier borrowers.3 Not
surprisingly, these riskier loans came at the price of higher interest rates to
compensate lenders for the increased risk that they undertook. But high
prices themselves are not the central problem. The root cause of the prob-
lem is that lenders hid these high prices and borrowers underappreciated
them. In the prime market, the traditional loan is a standardized, thirty-year
fixed-rate mortgage (FRM). Lenders could have covered the increased risk
of subprime loans by simply raising the interest rate on the traditional FRM.
They chose not to do this for a number of reasons. To uncover some of
those reasons, let’s look at the features of the common subprime mortgage
contract.
The subprime market boasted a broad variety of complex loans with
multidimensional pricing structures. Hybrid loans (which combine fixed
and variable rates), interest-only loans, and option-payment adjustable-rate
mortgages (ARMs)—each product type with its own multidimensional
design—were all common in the expanding subprime market. It should be
noted that many of these contractual designs were known in the prime
market since the early 1980s. But it was in the subprime market where they
first took center stage.4
Common subprime mortgage contracts exhibited the two design fea-
tures highlighted throughout this book; complexity and cost deferral. Let’s
consider cost deferral first. Again, an important note: All loan contracts, of
course, involve deferred costs. The loans we’re examining feature deferral
of costs beyond that which is necessarily implied by the nature of a loan.
3. See Demyanyk and Van Hemert, above note 1, at 1852, 1854 (tbl. 1); Center for Responsible
Lending, “Mortgage Lending Overview,” <http://www.responsiblelending.org/issues/
mortgage/>.
4. A note on terminology: The residential mortgage market is divided into the prime segment
and the nonprime segment. The nonprime segment is further divided into subprime (higher
risk) and Alt-A (lower risk), although the line between subprime and Alt-A is not always clear.
See below Part I.A. Many of the contractual design features studied in this chapter were com-
mon in both the subprime and Alt-A segments. For expositional convenience, I will sometimes
refer to these two segments together as “subprime”.
The traditional prime mortgage required a 20 percent down-pay-

ment, which implies a loan-to-value (LTV) ratio of no more than 80
percent. In the subprime market, in 2006, more than 40 percent of loans
had LTVs exceeding 90 percent. Focusing on purchase-money loans, in
2005, 2006, and the first half of 2007, the median subprime borrower put
no money down, borrowing 100 percent of the purchase price of the
house.
The schedule of payments on the loan itself exhibits the same deferred-
cost characteristic. Under the standard prime FRM, the borrower pays
the same dollar amount each month—a flat payment schedule. Under a
conventional ARM, where the monthly payment is calculated by adding
a fixed number of percentage points to a fluctuating index, the dollar
amount paid varies from month to month but without any systemic
trajectory. By way of contrast, the majority of subprime loans exhibited
an increasing payment schedule. They set a low interest rate for an intro-
ductory period—typically two years—and a higher interest rate for the
remaining term of the loan. Other subprime loans exhibited an even
steeper payment schedule. Interest-only loans and payment-option
ARMs allowed for zero or negative amortization during the introduc-
tory period, further increasing the step-up in the monthly payment after
the introductory period ended. A direct impact of an escalating-pay-
ments contract is often “payment shock,” which occurs when a rate-
reset leads to a significant increase in the monthly payment—sometimes
as much as 100 percent.
The second contractual design feature, common to most subprime
mortgages, is complexity. While the traditional FRM sets a single, constant
interest rate, the typical subprime mortgage includes multiple interest rates,
some of which are implicitly defined by nontrivial formulas that adjust
rates from one period to the next. The typical subprime loan also features
a host of fees, some applicable at different time periods during the loan
term, some contingent on various exogenous changes or on borrower
behavior. The numerous fees associated with a subprime loan fall under
two categories:
(1) Origination fees, including a credit check fee, an appraisal fee, a flood
certification fee, a tax certification fee, an escrow analysis fee, an under-
writing analysis fee, a document preparation fee, and separate fees for
sending emails, faxes, and courier mail; and
mortgage s 119
(2) Post-origination fees, including late fees, foreclosure fees, prepayment

penalties, and dispute-resolution or arbitration fees.
These fees can add up to thousands of dollars or up to 20 percent of the loan
amount. The prepayment option, of special importance in the subprime
market, further complicates the valuation of these contracts, as does the
(implicit) default option.
Because a borrower must choose among many different, complex prod-
ucts, each with a different set of multidimensional prices and features, the
complexity of the borrower’s decision is exponentially greater than the
already high level of complexity of a single contract.5
B. Contract Design Explained

What explains these contractual design features? To find out, let’s explore
possible rational-choice explanations, starting with the cost-deferral feature.
A common explanation for deferred-cost contracts is based on the afford-
ability argument. Many subprime borrowers, at the time they take out their
loans, are liquidity constrained: They can afford only a small down payment
and a small monthly payment. The catch, of course, is that a small down
payment and a small initial monthly payment eventually convert to higher
monthly payments when the initial rate resets to the post-introductory level.
Accordingly, the rationality of the affordability argument depends on the
borrower’s ability to either make the high future payment or to avoid it.
In this way, the argument then splits into two sub-arguments: the “make”
argument and the “avoid” argument.
The “make” argument holds that the borrowers will anticipate being
able to make the higher payment if they expect their income to increase
substantially by the end of the introductory period. Some subprime bor-
rowers rationally expected such a substantial increase in income; many
others did not.
The “avoid” argument posits that borrowers will be able to avoid the
higher payment if they expect to prepay the mortgage before the introduc-
tory period ends.The prepayment option depends on the availability of refi-
nance loans with attractive terms. Attractive refinancing options will be
5. “Truth in Lending,” 73 Fed. Reg. 44,522, 44,524–25 (July 30, 2008) (codified at 12 C.F.R. pt.
226) (“[P]roducts in the subprime market tend to be complex, both relative to the prime mar-
ket and in absolute terms. . . .”).
available if the borrower’s credit score improves, market interest rates fall, or
house prices increase. Some borrowers during the subprime mortgage boom
rationally expected that such positive realizations would enable them to refi-
nance their deferred-cost mortgage and avoid the high long-term costs. For
many other borrowers, these expectations were overly optimistic.
An alternative, rational-choice explanation portrays the deferred-cost
mortgage as an investment vehicle designed to facilitate speculation on real-
estate prices. According to this explanation, if house prices rise, the specula-
tor will be able to sell the house (or refinance) and pocket the difference
between the lower buy price and the higher sell price without ever paying
the high long-term cost of the deferred-cost loan. If house prices fall, the
explanation continues, the speculator will default on the mortgage—again
avoiding the high long-term cost. Of course, default is not a cost-free prop-
osition, but as long as the probability of a price increase is high enough, the
upside benefit will offset the downside risk. Some subprime borrowers were
surely speculators and benefited from this strategy. Many others, however,
were not—and did not.
Let’s turn to the second design feature; complexity and multidimension-
ality. We’ll start with the multiple, indirectly defined interest rates.6 The
index-driven rate adjustments of an ARM—further complicated by maxi-
mum adjustment caps—can be explained as a means to efficiently allocate
the risk of fluctuating interest rates between lenders and borrowers. This
explanation, however, was more powerful when interest-rate risk was shared
by the lender and borrower. During the subprime expansion, when securi-
tization was prevalent, this risk could have been—and sometimes was—
passed on to diversified investors.
Next, consider the proliferation of fees common in subprime mortgage
contracts. A rational-choice model can explain at least some of these fees.
Charging separate fees for separate services allows each borrower to pick and
choose from the offered services according to individual preferences. But this
applies only to optional services, not to the many non-optional yet separately
priced services, such as the credit check and document preparation. Another
explanation holds that the proliferation of fees reflects efficient risk-based pric-
ing. For example, delinquency imposes a cost on lenders. Late fees and fore-
closure fees allocate this cost to the delinquent borrowers. Absent such fees,
6. To the extent that interest-rate complexity is an artifact of the deferred-cost features, the pre-
ceding discussion applies here as well.
mortgage s 121
lenders would raise interest rates to compensate for the forgone fees, and non-
delinquent borrowers would bear a large share of the costs imposed by delin-
quent borrowers. Again, this explanation is plausible only for certain fees.
Rational-choice theories, then, explain some of the observed contractual
designs in some contexts. They do not provide a complete account. A
rational-choice model does not fully explain the prevalence of cost deferral
and the exceedingly high level of complexity. To fill this explanatory gap,
we’ll consider a behavioral-economics theory of the subprime mortgage
contract, applying the general framework of Chapter 1 to the mortgage world.
We’ll see that the design of subprime mortgage contracts can be explained as
a rational market response to the imperfect rationality of borrowers.
The cost-deferral feature can be traced back to borrowers’ myopia and
optimism bias. Myopic borrowers focus on the short-term dimensions of
the loan contract and pay insufficient attention to the long-term dimen-
sions. Optimistic borrowers underestimate the future cost of the deferred-cost
contract. They overestimate their future income and expect unrealistically
attractive refinance options. Or they overestimate the expected value of a
bet placed on the real-estate market, perhaps because they irrationally expect
that a 10 percent price increase last year will be replicated next year. If
myopic and optimistic borrowers focus on the short term and discount the
long term, lenders will offer deferred-cost contracts with low short-term
prices and high long-term prices.7
A similar argument explains the complexity of subprime mortgage con-
tracts. Imperfectly rational borrowers will not be able to effectively aggre-
gate multiple price and non-price dimensions and discern from them the
true total cost of the mortgage product. Inevitably, these borrowers will
focus on a few salient dimensions. If borrowers cannot process complex,
multidimensional contracts and thus ignore less salient price dimensions,
lenders will offer complex, multidimensional contracts, shifting much of the
loan’s cost to the less salient dimensions.8
7. See Ben S. Bernanke, Chairman, Bd. of Governors of the Fed. Reserve Sys., Speech at the
Women in Housing and Finance and Exchequer Club Joint Luncheon, Washington, D.C.:
“Financial Markets, the Economic Outlook, and Monetary Policy” (January 10, 2008), available
at <http://www.federalreserve.gov/newsevents/speech/bernanke20080110a.htm> (suggesting that
the ARM design responds to optimism about house prices).
8. See Edmund L. Andrews,“Fed and Regulators Shrugged as the Subprime Crisis Spread: Analy-
sis Finds Trail of Warnings on Loans,” New York Times, December 18, 2007, at A1 (quoting
Edward M. Gramlich, the former Federal Reserve governor, asking: “Why are the most risky
loan products sold to the least sophisticated borrowers . . . The question answers itself—the least
sophisticated borrowers are probably duped into taking these products.”).
While focusing on only one part of the subprime picture—the design of

subprime loan contracts—this chapter develops an alternative account of the
dynamics that led to the subprime crisis. One common account focuses on
unscrupulous lenders who pushed risky credit onto borrowers who were
incapable of repaying. Another common account focuses on irresponsible
borrowers who took out loans they could not repay. Both accounts capture
some of what was going on during the subprime boom, but both accounts
are incomplete. In many cases, borrowers were not reckless; they were
imperfectly rational. And in many cases lenders were not evil; they were
simply responding to a demand for financing driven by borrowers’ imper-
fect rationality.
This chapter highlights a demand-side market failure; imperfectly rational
borrowers “demanded” complex, deferred-cost loan contracts and lenders
met this demand. But the failures in the subprime mortgage market were
not limited to the demand side. A supply-side market failure explains why
lenders willingly catered to borrowers’ imperfectly rational demand—even
when the demanded product designs increased the default risk borne by
lenders.9
The main culprit in this supply-side failure was securitization—the
process of issuing securities backed by large pools of mortgage obligations.
Securitization created a host of agency problems as a series of agents—
intermediaries responsible for originating loans, pooling and packaging
them into mortgage-backed securities, and assessing the risk associated
with the different securities—stood between the principals (the investors
who ultimately funded the mortgage loans) and the borrowers. The com-
pensation of these agents/intermediaries was not designed to align their
interests with those of the principals/investors. Instead, their fees were
based on the quantity, not quality, of processed loans. Consequently, the
agents/intermediaries had strong incentives to increase the volume of
originations—even low-quality, high-risk ones—by promoting mortgage
products that, through high levels of complexity and cost deferral, created
the appearance of affordability.10
9. An immediate response is that lenders priced the increased risk. But this response is mislead-
ing. The evidence shows that subprime risks were not accurately priced.
10. See HUD, Report to Congress on the Root Causes of the Foreclosure Crisis, 31–2 (2010)
(hereinafter “HUD Report”).
mortgage s 123
Another factor: It is likely that even sophisticated investors and financial

intermediaries got caught up in the frenzy of the real-estate boom and
underestimated the risks associated with the mortgage products that they
were peddling.11 The multibillion dollar losses incurred by these sophisti-
cated players provide (at least suggestive) evidence that imperfect rationality
was not confined to the demand side of the subprime market.12 And there
is more direct evidence. For example, in an email message uncovered by an
SEC investigation, Angelo R. Mozilo, Countrywide’s CEO, wrote: “We
have no way, with any reasonable certainty, to assess the real risk of holding
these loans [Payment Option ARMS] on our balance sheet. . . .The bottom
line is that we are flying blind on how these loans will perform in a stressed
environment of higher unemployment, reduced values and slowing home
sales.”13
C. Welfare Implications
Back to the demand-side failure, which is the focus of this chapter: The
proposed behavioral-economics theory offers a more complete account of sub-
11. See HUD Report, above note 10, at 37 (concluding that investors overestimated the ability of
innovations in financial market instruments (e.g., CDOs and CDSs) to shield them from risk).
Much of this underestimation of risk harkens back to optimism about house prices. See, e.g.,
Kristopher S. Gerardi et al., Making Sense of the Subprime Crisis (2008) (Fed. Reserve Bank of
Boston, Public Policy Discussion Paper No. 09–1), 1, available at <http://www.bos.frb.org/eco-
nomic/ppdp/2009/ppdp0901.htm> (finding that analysts in 2005 understood the risks of a
steep decline in house prices but believed that the probability of such a decline was very low);
Julio Rotemberg, Subprime Meltdown: American Housing and Global Financial Turmoil (Harvard
Business School, 2008) 1 (quoting a letter that Fannie Mae CEO Franklin Raines sent to
shareholders in 2001: “Housing is a safe, leveraged investment—the only leveraged investment
available to most families—and it is one of the best returning investments to make . . . Homes
will continue to appreciate in value. Home values are expected to rise even faster in this
decade than in the 1990s.”).
12. See Jennifer E. Bethel, Allen Ferrell, and Gang Hu, “Legal and Economic Issues in Litigation
Arising from the 2007–2008 Credit Crisis” 21, 81 tbl. 2 (2008) (Harvard Law and Econ. Discus-
sion Paper No. 212), available at <http://ssrn.com/abstract=1096582> (summarizing the tens
of billions of dollars worth of subprime-related write-offs by banks; citing an estimate of $150
billion in writedowns as of February 2008 and a forecast that this amount will more than
double); Press Release, Standard and Poor’s, “Subprime Write-Downs Could Reach $285 Bil-
lion, but Are Likely Past the Halfway Mark” (March 13, 2008), available at <http://www2.
standardandpoors.com/portal/site/sp/en/us/page.article/4,5,5,1,1204834027864.html> (dis-
cussing Standard and Poor’s increased estimate of writedowns at $285 billion, up from $265
billion earlier in the year). These losses do not provide conclusive evidence that sophisticated
players made mistakes; they could be the realization of the large (!) downside risk in an (ex
ante) rational bet.
13. See Gretchen Morgenson, “S.E.C. Accuses Countrywide’s Ex-Chief of Fraud,” New York
Times, June 4, 2009 (quoting from Mozilo’s email).
prime market dynamics and of how these dynamics shaped the design of
subprime loan contracts. These contractual design features have significant
welfare implications, especially when understood as a market response to
the imperfect rationality of borrowers. First, excessive complexity prevents
effective comparison-shopping and thus hinders competition. Second,
deferred-cost features are correlated with increased levels of delinquency
and foreclosure, which impose significant costs not only on borrowers but
also on surrounding communities, lenders, loan purchasers, and the econ-
omy at large.Third, excessively complex deferred-cost contracts have adverse
distributive consequences, disproportionally burdening financially weaker—
often minority—borrowers. Finally, loading a loan’s cost onto less salient or
underappreciated price dimensions artificially inflates the demand for mort-
gage financing and, indirectly, for residential real estate.The proposed theory
thus establishes a causal link between contractual design and the subprime
expansion and real-estate boom. Accordingly, the subprime meltdown that
followed this expansion can also be attributed, at least in part, to the identi-
fied contractual design features.14
Importantly, the identified contractual design features and the welfare
costs associated with them are not necessarily the result of less-than-vigor-
ous competition in the subprime market. Enhanced competition could
even make these design features more pervasive. If borrowers focus on the
short term and discount the long term, competition might force lenders to
offer deferred-cost contracts. And if borrowers faced with complex, multi-
dimensional contracts ignore less salient price dimensions, competition
might force lenders to offer complex, multidimensional contracts and to
shift much of the loan’s cost to the less salient price dimensions.
On the other hand, competition may create incentives for lenders to
educate consumers and reduce the biases and misperceptions that give rise
to excessive complexity and cost deferral. Measures designed to ensure
robust competition in the subprime mortgage market, while desirable,
should not be relied upon to solve the behavioral market failure.
14. See Section V.B. below. While contractual design contributed to the subprime expansion,
there are other factors that likely played a more central role in generating the subprime
expansion, including (1) the advent of new technology that enabled efficient risk-based pric-
ing, and (2) the increase in the supply (or availability) of funds brought about by securitization
and the global saving glut.
mortgage s 125
D. Policy Implications
The subprime crisis has spurred a plethora of reforms and reform propos-
als.15 In this chapter, we’ll focus on disclosure regulation—specifically, the
Annual Percentage Rate (APR) disclosure, the traditional centerpiece of
the mortgage disclosure regime.
The importance of the APR, and of total-cost-of-credit measures more
generally, has been reaffirmed by the new laws and regulations. The Mort-
gage Reform and Anti-Predatory Lending Act, enacted as Title XIV of the
Dodd–Frank Wall Street Reform and Consumer Protection Act of 2010,
requires disclosure of total-cost information. And the new disclosure forms
being developed by the Consumer Financial Protection Bureau (CFPB),
the result of a direct Congressional mandate in the Dodd–Frank Act, secure
an important place for the APR.16 The APR disclosure has the potential to
undo the adverse effects of imperfect rationality, including the identified
contractual design features and the welfare costs they impose.
The APR disclosure was the most important innovation of the Truth in
Lending Act (TILA) of 1968.17 A normalized total-cost-of-credit measure, the
APR was designed to assist borrowers in comparing different loan products.
In theory, the APR should solve—or at least mitigate—both the complexity
and cost-deferral problems. Complexity and multidimensionality pose a prob-
lem if they hide the true cost of the loan. The APR responds to this concern
by folding the multiple price dimensions into a single measure. The APR
should similarly help short-sighted borrowers grasp the full cost of deferred-
cost loans, as the APR calculation assigns proper weight to the long-term
price dimensions. Moreover, since the APR—in theory—strips away any
competitive advantage of excessive complexity and cost deferral, lenders will
have no reason to offer loan contracts with these design features.
15. Major new laws and regulations include (1) the Mortgage Reform and Anti-Predatory Lend-
ing Act, enacted as Title XIV of the Dodd–Frank Wall Street Reform and Consumer Protec-
tion Act of 2010, Pub. L. 111–203 (hereinafter “Dodd–Frank Act’), and (2) the new set of
regulations governing mortgage lending issued by the Federal Reserve Board (FRB) in July
2008–“Truth in Lending,” 73 Fed. Reg. 44,522, 44,524–25 (July 30, 2008) (codified at 12
C.F.R. pt. 226).
16. See Dodd–Frank Act, Sec. 1419 (requiring disclosure of total cost information) and Sec.
1032(f) (directing the CFPB to develop a new disclosure form). See also CFPB,“Know Before
You Owe Initiative,” <http://www.consumerfinance.gov/knowbeforeyouowe/>.
17. Truth in Lending Act, Pub. L. No. 90–321, § 107, 82 Stat. 146, 149 (1968) (codified as amended
at 15 U.S.C. § 1606 (2006)) (defining the APR); Truth in Lending Act, Pub. L. No. 90–321,
§§ 121–31, 82 Stat. 146, 152–7 (1968) (codified as amended at 15 U.S.C. §§ 1631–49 (2006))
(requiring disclosure of the APR).
The APR can solve these problems, but only if it lives up to the expecta-
tions of the Congress that enacted it, namely, if it provides a timely, true
measure of the total cost of credit and borrowers rely on it in choosing
among different loan products. In the run-up to the subprime crisis, the
APR disclosure did not live up to these expectations. Why? Three reasons.
First, the APR disclosure often came too late to be useful for comparison-
shopping. Second, the APR did not, and still does not, measure the total cost
of credit. This is because numerous fees paid by mortgage borrowers are
excluded from the regulatory definition of a “finance charge” and are thus
ignored in the APR calculation.
Third, the APR calculation assumed, and still assumes, that the borrower
will hold the loan for the nominal loan period, typically 30 years. The actual
duration of a mortgage loan is, however, closer to five years on average in the
subprime market. Most borrowers refinance and prepay (or default) long
before the thirty-year mark. By ignoring the possibility of prepayment (and
default) the APR disclosure fails to reflect the true total cost of the loan.This
distortion was especially pronounced during the recent subprime expansion,
when for many loans the prepayment option constituted a substantial value
component. When a borrower expects to prepay a deferred-cost loan by the
end of the low-rate introductory period, it makes little sense for this bor-
rower to rely on an APR that presumes continued payments at the high
post-introductory rate. Since the APR disclosure often came too late and did
not reflect the true cost of credit, borrowers stopped relying on the APR as
the main tool for comparison shopping among loan products. This dimin-
ished its power as an effective antidote to imperfect rationality.
Recent reforms and existing reform proposals address some of the short-
comings of the APR disclosure. The timing-of-disclosure problem was
addressed and partially solved by the Federal Reserve Board’s (FRB’s) new
mortgage regulations and by the recently enacted Housing and Economic
Recovery Act.18 These reforms are commendable, but more should be done.
In the most recent and most comprehensive proposal to create a more
inclusive APR, Elizabeth Renuart and Diane Thompson are advocating a
broader definition of a “finance charge,” one that would cover all, or most,
18. See Truth in Lending, 73 Fed. Reg. at 44,524 (“The final rule requires creditors to provide
transaction-specific mortgage loan disclosures such as the APR and payment schedule for all
home-secured, closed-end loans no later than three business days after application, and before
the consumer pays any fee except a reasonable fee for the review of the consumer’s credit
history.”); Housing and Economic Recovery Act of 2008 § 2502(a).
mortgage s 127
of the costs paid by borrowers.19 The analysis in this chapter supports the
spirit of the Renuart–Thompson proposal while also recognizing that a
comprehensive cost-benefit analysis may justify keeping certain price
dimensions outside the scope of the “finance charge” definition.
Recent reforms and existing reform proposals do not address the exclu-
sion of the prepayment option (nor the default option) from the APR
definition.We’ll explore ways the APR calculation would have to be adjusted
to incorporate the prepayment option. There are costs involved, of course,
and policymakers will need to carefully weigh these costs against the poten-
tially substantial benefits of an APR that accounts for the prepayment
option. But if borrowers ignored the traditional APR figure because it
excluded the prepayment option, they should embrace an APR that incor-
porates that option. And as the APR reclaims its rightful position at the
forefront of the mortgage disclosure regime, borrowers—and society—will
again benefit from the APR’s unique ability to undo the adverse effects of
imperfect rationality.
While this chapter focuses on the subprime mortgage market, much of
the analysis applies to the other segments of the residential mortgage mar-
ket—the Alt-A segment and even to the prime segment. There, too, highly
complex, deferred-cost contracts began to appear in increasing numbers,
alongside the traditional FRM. In fact, the most extreme forms of cost
deferral—the interest-only and payment-option mortgages—were more
common in the Alt-A and prime segments than in the subprime segment.
Moreover, it was in the Alt-A and prime segments where introductory rates
were substantially below the fully indexed market rate. While the crisis
began with subprime, it did not end there. Defaults and foreclosures have
appeared in substantial numbers in the Alt-A and even prime markets.20
19. See Elizabeth Renuart and Diane E. Thompson, “The Truth, the Whole Truth, and Nothing
but the Truth: Fulfilling the Promise of Truth in Lending” Yale J. Reg. 25 (2008), 181. Renuart
and Thompson, however, are not the first to recognize that the APR is not sufficiently inclu-
sive, nor are they the first to propose a more inclusive APR. See U.S. Dep’t of Hous. and
Urban Dev. and U.S. Dep’t of the Treasury, Curbing Predatory Home Mortgage Lending (2000)
69, available at <http://www.huduser.org/publications/hsgfin/curbing.html> (hereinafter
“HUD-Treasury Report”) (proposing that the law be amended “to require that the full costs
of credit be included in the APR”); William N. Eskridge, Jr., “One Hundred Years of Inepti-
tude: The Need for Mortgage Rules Consonant with the Economic and Psychological
Dynamics of the Home Sale and Loan Transaction”, Va. L. Rev., 70 (1984), 1083, 1166 (pro-
posing, over twenty years ago, a more inclusive APR).
20. On cost deferral in the Alt-A market—see Christopher L. Cagan, First Am. CoreLogic, Inc.,
Mortgage Payment Reset: The Issue and the Impact (2007) 2. On defaults and foreclosures in the
Alt-A and Prime markets—see Stan J. Liebowitz, “Anatomy of a Train Wreck: Causes of the
This chapter proceeds as follows:

• Part I provides some background on the subprime mortgage market.
• Part II describes the central design features of subprime mortgage
contracts.
• Part III evaluates the rational-choice explanations for the identified con-
tractual design features, emphasizing the limits of these rational-choice
theories.
• Part IV develops an alternative, behavioral-economics theory that fills
the explanatory gap left by the rational-choice accounts.
• Part V describes the welfare costs of the identified contractual design
features.
• Part VI considers policy implications.
I. The Subprime Mortgage Market
A. Defining Subprime
Subprime mortgage loans are generally sold to riskier borrowers.These high-
risk loans come at a high price (a high interest rate), as compared to the
less-risky prime loans that come at a lower price (a lower interest rate). But
this definition and the difference between prime and subprime lending
establishes a misleading dichotomy. The risks and, accordingly, loan prices
vary along a continuum. Still, because the mortgage industry itself follows
this rough categorization, as do policymakers, it is helpful to focus on a sub-
set of high-risk, high-price loans even if the line that divides this category of
loans from the lower-risk, lower-price category is arbitrary and blurry.
While the boundaries of the subprime segment may be fuzzy, the indus-
try, researchers, and regulators all use the same parameters to classify a loan
as subprime. According to one rough division, borrowers with FICO
scores—a common measure of creditworthiness—lower than 620 are con-
sidered subprime borrowers. Of course, a borrower’s FICO score is only
Mortgage Meltdown,” in Holcombe, R.G., and Powell, B. (eds.), Housing America: Building
Out of a Crisis (The Independent Institute, 2009) (explaining that ARM defaults and foreclos-
ures are as prevalent in the prime market as in the subprime market).
mortgage s 129
one of several factors determining risk level. Thus, industry participants

consider additional risk factors, such as the LTV ratio, when classifying a
loan as subprime.21 Moving from risk factors to price, a common subprime
threshold is a loan APR that is three or more points above the treasury rate
for a security of the same maturity. The three-point threshold defines
“higher-priced loans” under the Home Mortgage Disclosure Act (HMDA).
In its new subprime mortgage regulations, the FRB adopted a slightly dif-
ferent definition of “higher-priced mortgage loans,” setting the threshold
APR at 1.5 points above the “average prime offer rate.”22
B. Subprime Mortgage Loans: The Numbers

The subprime mortgage market grew substantially from the early 2000s to
2006. In 2001, approximately 985,000 first-lien subprime loans were origi-
nated, while in 2006 that number was approximately 2,780,000 and repre-
sented over 20 percent of total loan-origination volume. According to the
Congressional Budget Office (CBO), subprime mortgages at the end of
2006 accounted for 13 percent of all outstanding home mortgage loans.The
Alt-A market—encompassing “medium risk” loans between subprime and
prime—also experienced significant growth, expanding from 2 percent of
total originations in 2003 to 13 percent of originations in 2006.23
21. Credit Suisse, Mortgage Liquidity du Jour: Underestimated No More (2007) 13, 21 (hereinafter
“Credit Suisse Report”); Kristopher Gerardi, Adam Hale Shapiro, and Paul S. Willen, “Sub-
prime Outcomes: Risky Mortgages, Homeownership Experiences, and Foreclosures” (2007)
5–7 (Fed. Reserve Bank of Boston, Working Paper No. 07-15), available at <http://www.bos.frb.
org/economic/wp/wp2007/wp0715.htm>.
22. For the older HMDA definition—see Michael LaCour-Little, “Economic Factors Affecting
Home Mortgage Disclosure Act Reporting”, J. Real Est. Res., 29 (2007), 479, 506 n.3. For the
new definition—see Truth in Lending, 73 Fed. Reg. 44,522, 44,531–32 (codified at 12 C.F.R.
pt. 226) (stating that “[t]he definition of ‘higher-priced mortgage loans’ appears in § 226.35(a)”
and that the average prime offer rate is derived from the Freddie Mac Primary Mortgage
Market Survey®).
23. See Demyanyk and Van Hemert, above note 1, at 1853, 1854 (tbl. 1). (The authors’ data include
452,000 loans in 2001 and 1,772,000 loans in 2006. These data cover approximately 85 percent
of securitized subprime loans. In 2001, 54 percent of subprime loans were securitized, implying
a total of 452,000/(0.85 * 0.54) = 984,749. In 2006, 75 percent of subprime loans were securi-
tized, implying a total of 1,772,000/(0.85 * 0.75) = 2,779,608.); CBO Outlook, above note 2, at
23–4 (reporting the 20 percent and 13 percent figures); Christopher J. Mayer, Karen M. Pence,
and Shane M. Sherlund, “The Rise in Mortgage Defaults” 3 (2008) (Bd. of Governors of the Fed.
Reserve Sys., Fin. and Econ. Discussion Series Paper No. 2008–59) (recording LP data showing a rise
in subprime originations from 1.1 million in 2003 to 1.9 million in 2005). On the Alt-A mar-
ket—see Truth in Lending, 73 Fed. Reg. at 44,533; see also Mayer et al., id., at 3 (explaining that
“Alt-A originations grew . . . from 304,000 in 2003 to 1.1 million in 2005”).
The average size of a subprime loan has also increased. In 2006, the aver-
age size of a first-lien subprime loan was $212,000, up from $126,000 in
2001. In terms of loan purpose, in 2006, 42.4 percent of first-lien subprime
loans were purchase loans; 57.6 percent were refinance loans. The average
subprime borrower had a debt-to-income ratio of approximately 40 percent
and a FICO score of 618.1. The median subprime borrower had a FICO
score of 620. The median Alt-A borrower had a FICO score of 705.24
C. Market Structure
1. Participants
Traditionally, a single entity, commonly the neighborhood bank, was the
only party involved in the mortgage transaction (other than the borrower,
of course). The bank would originate the loan, provide the funds for the
loan, and service the loan. In the modern mortgage market, however, these
various roles—origination, financing, and servicing—are often performed
by different entities.25 Let’s focus on the parties involved in origination and
financing, since they exert the most influence on the design of the mort-
gage contract.
In the subprime and Alt-A markets, mortgages were originated primarily
by depository institutions (banks or bank subsidiaries and affiliates) and by
mortgage companies, with the bulk of loan volume originated by mortgage
companies.Another important group of participants in the mortgage origin-
ation process, brokers, accounted for 58 percent of total origination activity
in 2006.26
24. See Demyanyk and Van Hemert, above note 1, at 1854 (tbl. 1); Mayer et al., above note 23, at
6; Michael Fratantoni et al., Mortgage Bankers Ass’n, The Residential Mortgage Market and Its
Economic Context in 2007 (MBA Research Monograph Series, 2007) at 24.
25. See Henry M. Paulson, Jr., U.S. Sec’y of the Treasury, “Remarks on Current Housing and
Mortgage Market Developments at the Georgetown University Law Center” (October 16,
2007), available at <http://www.treasury.gov/press/releases/hp612.htm> (“A mortgage loan
is likely to be originated, serviced, and owned by three different entities. Originators often
sell mortgages to securitizers who package them into mortgage-backed securities, which are
then divided and sold again to a global network of investors.”).
26. U.S. Gov’t Accountability Office, Report to the Chairman, Subcommittee on Housing and Trans-
portation, Committee on Banking, Housing, and Urban Affairs, U.S. Senate, GAO-06–1021, Alter-
native Mortgage Products: Impact on Defaults Remains Unclear, but Disclosure of Risks to Borrowers Could
Be Improved (2006) 7 (hereinafter “GAO AMP Report”); Robert B. Avery, Kenneth P. Brevoort,
and Glenn B. Canner,“Opportunities and Issues in Using HMDA Data”, J. Real Est. Res. 29 (2007),
351, 353; Press Release, Access Mortgage Research and Consulting, Inc., “New Broker Research
Published” (August 17, 2007), available at <http://accessmtgresearch.com/?p=40>.
mortgage s 131
Traditionally, depository institutions originated loans and funded them

with the deposits they held. During the subprime expansion, origination
volume shifted to mortgage companies that had no independent means to
fund the originated loans. These mortgage companies, and increasingly also
depository institutions, sold the loans that they originated to Wall Street
investment banks. The banks pooled the loans, carved up the expected cash
flows, and converted these cash flows into bonds that were secured by the
mortgages. At the peak of the subprime expansion, most mortgages were
financed through this process of securitization. As a result, the “owners” of
the loans were the investors who purchased shares in these Mortgage (or
Asset) Backed Securities (MBSs or ABSs).27
Loan originators have direct control over the design of the mortgage
contract. Investment banks and their clients also influence the design of
mortgage contracts because the demand for MBSs—and thus the price that
investment banks are willing to pay the originators for the loans—depends
on the contractual design.
2. Competition
Competition in a market can also affect the design of the products and con-
tracts sold in this market. In the years before the subprime crisis, the loan
origination market appeared to be fairly competitive. In 2006, the top fif-
teen subprime lenders accounted for 80.5 percent of the market, with no
lender holding more than a 13-percent share. The Department of Housing
and Urban Development’s (HUD) list of lenders that specialize in subprime
lending named 210 lenders (although not all of these lenders offer loans
nationally). Barriers to entry in this industry were substantially lowered
with the growth in securitization, which enables entry by new, small lend-
ers. The internet enhanced competition by reducing shopping costs. For
these reasons, the FRB characterized this market as competitive.28
27. See, e.g., Kathleen C. Engel and Patricia A. McCoy, “Turning a Blind Eye:Wall Street Finance
of Predatory Lending”, Fordham L. Rev. 75 (2007), 2039, 2045. On securitization rates—see
Credit Suisse Report, above note 21, at 11 (finding 75 percent securitization rate); Demyanyk
and Van Hemert, above note 1, at 1853 (n. 6) (reporting securitization rates of 76 percent and
75 percent in 2005 and 2006, respectively).
28. On market shares—see Credit Suisse Report, above note 21, at 22. See also 2 Market Share Reporter:
An Annual Compilation of Reported Market Share Data on Companies, Products, and Services: 2008
(Robert S. Lazich, ed., 2008) at 704–5 (reporting that the top ten lenders commanded less than
58.8 percent of the market with no single lender controlling more than 8.3 percent of the market,
Nevertheless, because many consumers engage in limited shopping, sev-

eral observers have expressed concerns about the level of competition in the
subprime market. As will be explained in Section V.1, the increasing com-
plexity of mortgage products renders comparison shopping more difficult
and limits the efficacy of the shopping that does take place. This limited
shopping may be a rational response to its reduced efficacy. The result:
imperfect information and imperfect competition.
HUD’s amendments to its Real Estate Settlement Procedures Act
(RESPA) regulations were intended to enhance competition in the mort-
gage market. Two studies—one by the Government Accountability Office
(GAO), the other by the Federal Trade Commission (FTC) and the Depart-
ment of Justice (DOJ)—expressed concern about the level of competition
in the real estate brokerage industry, which, as explained above, plays an
important role in the loan origination process.29
based on a conservative combination of the two sources cited in Market Share Reporter).
On HMDA’s list of 210 subprime lenders—see Randall M. Scheessele, “HUD Subprime
and Manufactured Home Lender List,” HUD User, March 16, 2007, <http://www.huduser
.org/datasets/manu.html> (describing the 2005 list). Many other lenders, while not spe-
cializing in subprime lending, also offer subprime loans. See Avery et al., above note 41, at
353 (noting that there were 8,850 Home Mortgage Disclosure Act (HMDA) reporting
institutions in 2005). On the role of securitization in reducing barriers to entry—see Engel
and McCoy, above note 27, at 204. For an indication of the role played by the internet in
enhancing competition—see, e.g., LendingTree.com, Lender Ratings, <http://www
.lendingtree.com/stm3/lenders/scorecard.asp>. For the FRB’s assessment—see Truth in
Lending, 73 Fed. Reg. 1672, 1674 (proposed January 9, 2008) (codified at 12 C.F.R. pt. 226)
(“Underwriting standards loosened in large parts of the mortgage market in recent years as
lenders—particularly nondepository institutions, many of which have since ceased to
exist—competed more aggressively for market share.”).
29. On the concern of commentators about limited shopping—see Eskridge, above note 19; Marsha
J. Courchane, Brian J. Surette, and Peter M. Zorn, “Subprime Borrowers: Mortgage Transitions
and Outcomes”, J. Real Est. Fin. and Econ. 29 (2004), 365, 371–2; Lauren E.Willis, “Decisionmak-
ing and the Limits of Disclosure: The Problem of Predatory Lending: Price”, Md. L. Rev. 65
(2006), 707, 749. The limits of advertising in the subprime market further increase the cost of
comparison shopping. See Truth in Lending, 73 Fed. Reg. 44,522, 44,524 (July 30, 2008) (codified
at 12 C.F.R. pt. 226) (“[P]rice information for the subprime market is not widely and readily
available to consumers. A consumer reading a newspaper, telephoning brokers or lenders, or
searching the Internet can easily obtain current prime interest-rate quotes for free. In contrast,
subprime rates, which can vary significantly based on the individual borrower’s risk profile, are
not broadly advertised and are usually obtainable only after application and paying a fee.”). On
HUD’s RESPA Amendments and the motivation for these amendments—see “Real Estate Set-
tlement Procedures Act (RESPA): Rule to Simplify and Improve the Process of Obtaining Mort-
gages and Reduce Consumer Settlement Costs,” 73 Fed. Reg. 68,204, 68,207 (November 17,
2008) (codified at 24 C.F.R. pts. 203, 3500) (describing “important changes that should increase
consumer understanding and competition in the mortgage marketplace”). On the limited com-
petition among brokers—see U.S. Gov’t Accountability Office, Report to the Committee on
Financial Services, House of Representatives, “Real Estate Brokerage: Factors That May Affect
Price Competition” GAO-05-947 (2005); U.S. Dep’t of Justice and Fed.Trade Comm’n,“Competi-
mortgage s 133
As already noted, contractual design is not determined solely by the loan

originator. Thus, competition (or lack thereof ) in other markets may have
influenced the design of mortgage contracts. In particular, securitization
enhanced competition in the loan-origination market but simultaneously
transferred some control over contractual design away from the originators
and into the hands of securitizers.The securitization market appears to have
been relatively competitive. In 2007, the top 10 securitizers—Lehman
Brothers, Bear Stearns, Morgan Stanley, JP Morgan, Credit Suisse, Bank of
America Securities, Deutsche Bank, Royal Bank of Scotland Group, Merrill
Lynch, and Goldman Sachs—controlled 73.4 percent of the market, with
no single bank controlling more than 10.8 percent of the market.30
D. Regulatory Scheme
Before the financial crisis, regulatory authority over mortgage lending was
divided between the federal and state levels and among several regulators at the
federal level. Federal banking agencies—the FRB, the Office of the Comptrol-
ler of the Currency (OCC), the Office of Thrift Supervision (OTS), the Fed-
eral Deposit Insurance Corporation (FDIC), and the National Credit Union
Administration (NCUA)—regulated depository institutions.The Federal Trade
Commission Improvements Act of 1980 authorized the Federal Reserve to
identify unfair or deceptive acts or practices by banks and to issue regulations
prohibiting them. Moreover, the federal banking agencies could use Section 8
of the Federal Deposit Insurance Act to prevent unfair or deceptive acts or
practices under Section 5 of the Federal Trade Commission Act, regardless of
whether there was an FRB regulation defining the particular act or practice as
unfair or deceptive. Focusing on high-priced mortgage loans, the Home Own-
ership and Equity Protection Act (HOEPA) granted the FRB broad powers to
police unfair or deceptive lending practices.The FRB also promulgated disclos-
ure regulations under TILA. Non-depository institutions—that is, nonbanks,
including mortgage companies, brokers, and advertisers—fell under the juris-
diction of the FTC.The FTC enforced TILA, HOEPA and § 5 of the Federal
Trade Commission Act.31
tion in the Real Estate Brokerage Industry: A Report by the Federal Trade Commission and the U.S.
Department of Justice” (2007), available at <http://www.ftc.gov/reports/realestate/V050015.pdf>.
30. See Bethel et al., above note 12, at 81 tbl. 2.
31. On the regulation of unfair or deceptive acts or practices by the banking agencies—see 15
U.S.C. §§ 57b-1 to -4 (2006) (FRB authority to issue regulations); Comptroller of the Cur-
rency, Administrator of National Banks, “Guidance on Unfair or Deceptive Acts or
Practices,”Advisory Letter No. AL 2002–3 (March 22, 2002), available at <http://www.occ.treas
At the state level, mini-FTC statutes prohibited unfair and deceptive acts
and practices. Likewise, mini-HOEPA statutes, as well as other statutes,
banned or restricted specific practices, such as prepayment penalties and
balloon clauses. There was, however, substantial variation in the scope and
enforcement of state-level laws. Because some states clearly went further
than federal regulators in their attempts to protect borrowers, heated
preemption battles erupted, especially with the OCC and OTS. These bat-
tles were often won by the federal regulators.32
This regulatory landscape was substantially altered by the Dodd–Frank
Wall Street Reform and Consumer Protection Act of 2010. Congress rec-
ognized that when authority and responsibility for protecting borrowers is
divided between multiple federal agencies, coordination problems and reg-
ulatory competition gone awry leave borrowers inadequately protected.
Moreover, the old regulatory scheme exhibited a fundamental mismatch:
banking agencies that care more about the safety and soundness of financial
institutions and less about consumer protection were given the authority
to protect consumers, while the FTC—the one federal agency with a pri-
mary consumer protection mission—was denied authority over depository
institutions.33
.gov/ftp/advisory/2002-3.doc> (banking agencies’ enforcement authority); Truth in Lend-

ing, 73 Fed. Reg. 44,522, 44,527 (July 30, 2008) (codified at 12 C.F.R. pt. 226) (FRB authority
with respect to high-priced loans). On disclosure regulation—see 12 C.F.R., Part 226, Sub-
part C (Regulation Z). Additional disclosure regulations were promulgated by HUD under
RESPA, which governs the loan-closing process. See 24 C.F.R. pts. 203, 3500. On the FTC’s
authority with respect to non-depository institutions—see Letter from Donald S. Clark,
Sec’y, U.S. Fed. Trade Comm’n, to Jennifer L. Johnson, Sec’y, Bd. of Governors of the Fed.
Reserve Sys.1 (September 14,2006),available at <http://www.federalreserve.gov/SECRS/2006/
November/20061121/OP-1253/OP-1253_53_1.pdf> (hereinafter “FTC Comment”).
32. On state-level regulation—see Raphael W. Bostic et al., “State and Local Anti-Predatory
Lending Laws: The Effect of Legal Enforcement Mechanisms”, J. Econ. and Bus. 60 (2008), 47;
Anthony Pennington-Cross and Giang Ho,“The Termination of Subprime Hybrid and Fixed
Rate Mortgages” (2006) (Fed. Reserve Bank of St. Louis, Research Div., Working Paper No. 2006-
042A) 8–9, available at <http://research.stlouisfed.org/wp/2006/2006-042.pdf> Ctr. for
Responsible Lending, CRL State Legislative Scorecard: Predatory Mortgage Lending,
<http://www.responsiblelending.org/issues/mortgage/statelaws.html>. On the tension
between state-level and federal-level regulation and the preemption battles—see Oren Bar-
Gill and Elizabeth Warren, “Making Credit Safer” U. Penn. L. Rev. 157, (2008) 1, 79–83; Kurt
Eggert, “Limiting Abuse and Opportunism by Mortgage Servicers”, Housing Pol’y Debate, 15
(2007) 753, 774–5; Julia Patterson Forrester,“Still Mortgaging the American Dream: Predatory
Lending, Preemption, and Federally Supported Lenders”, U. Cin. L. Rev., 74 (2006), 1303;
Christopher L. Peterson,“Preemption, Agency Cost Theory, and Predatory Lending by Bank-
ing Agents: Are Federal Regulators Biting Off More Than They Can Chew?”, Am. U. L. Rev.,
56 (2007), 515 (2007).
33. See Bar-Gill and Warren, above note 32.
mortgage s 135
In response to these systemic failures, the Dodd–Frank Act created the Con-
sumer Financial Protection Bureau (CFPB) and entrusted it with broad author-
ity over providers of consumer credit. In essence, Congress took the consumer
protection powers that were dispersed across multiple agencies and gave them
to the CFPB, along with some new powers. Congress also sought to restore
some of the power taken from the states in the preemption battles.34
E. Summary
The subprime mortgage market experienced significant growth between
2000 and 2006. This rapid growth slowed dramatically in 2006, and when
the subprime crisis erupted in 2007, the market basically shut down.35 Still,
the analysis that follows is more than a historical account of a market that
no longer exists. While few new subprime loans are being originated, many
subprime loans are still outstanding.The goal of this chapter is to contribute
to a fuller assessment of the welfare costs that have been—and continue to
be—generated by these loans. The analysis will also point to policy reforms
that can prevent a second subprime crisis when subprime lending resumes.
Also, the results of our analysis will be relevant beyond subprime, as loan
contracts in other segments of the mortgage market share certain design
features with subprime contracts. Finally, an analysis of the subprime market
holds important lessons on the interaction between market forces and bor-
rower psychology—lessons applicable to other consumer credit markets and
even to noncredit markets.
II. The Subprime Mortgage Contract

The traditional, prime mortgage contract is a relatively simple, fixed-rate,
thirty-year loan for 80 percent or less of the home price (that is, requiring
a down payment of at least 20 percent). The typical subprime mortgage
contract was very different from this traditional benchmark. In this Part,
we’ll look at the two main design features that distinguish the common
34. See Dodd–Frank Act, Title X.

35. See, e.g., Ben S. Bernanke, Chairman, Bd. of Governors of the Fed. Reserve Sys., Testimony
before the Committee on the Budget, U.S. House of Representatives: The Economic Out-
look (January 17, 2008), available at <http://www.federalreserve.gov/newsevents/testimony/
bernanke20080117a.htm> (noting the “virtual shutdown of the subprime mortgage market”).
subprime mortgage contract from the traditional prime FRM: deferred

costs and a high level of complexity.
A. Deferred Costs
The common subprime loan defers costs via three contractual design fea-
tures: small down payments and high LTVs, escalating payments, and pre-
payment penalties.
1. Small Down Payments and High LTVs

The down payment, while not a component of the loan contract, is a com-
ponent of the payment stream that home buyers face. This payment stream
consists of a “time zero” payment, the down payment, and the payment
schedule specified in the loan contract. This broader, payment-stream per-
spective is helpful for two reasons. First, from the buyer’s perspective, it
makes little difference if a payment is made to the seller or to the lender.
Second, in many cases, a close (formal or informal) relationship between the
seller and the lender allows payment shifting between these two parties.36
One way to defer the costs associated with a home purchase is to reduce
the down payment. Indeed, the amount of the average down payment
declined during the subprime expansion. Traditionally, a home buyer was
required to make a down payment equal to at least 20 percent of the pur-
chase price. In 2005 and 2006, the median subprime home buyer put no
money down, borrowing 100 percent of the purchase price of the house.
Down payments were a bit higher in the Alt-A market, with a median value
of 5 percent in 2006.37
36. See Eskridge, above note 19, at 1124–7.

37. On traditional down payment requirements—see FTC Comment, above note 31, at 5. On
down payment requirements in 2005–06 in the subprime market—see Mayer et al., above
note 23, at 33 tbl. 2B; see also FTC Comment, above note 31, at 10 n. 45 (indicating that, in
the few years prior to 2005, over 40 percent of first-time home buyers did not make any
down payment at all); Gerardi et al., above note 21, at 44 tbl. 2 (finding—using the HUD-list
definition of “subprime”and Massachusetts data—that the average LTV of an initial-purchase
subprime loan rose from 0.76 in 1988 to 0.84 in 2007 and that the median LTV rose from 0.80
in 1988 to 0.90 in 2007); Amy Hoak, “100 Percent More Difficult: First-Time Home Buy-
ers Struggle to Find Down-Payment Money,” MarketWatch, March 9, 2008, <http://www.
marketwatch.com/news/story/first-time-home-buyers-struggle-find/story.aspx?guid
=%7B4BF19BC0-C4EE-4107-ACFC-F6524E878D5A%7D)> (stating that for the period
between July 2006 and June 2007, the National Association of Realtors estimated that 45
percent of first-time home buyers opted for 100 percent financing). On down payment
requirements in 2006 in the Alt-A market—see Mayer et al., above note 23, at 33 tbl. 2B.
mortgage s 137
The flip side of the down payment is the LTV ratio. In a purchase loan, a 10
percent down payment is equivalent to a 90 percent LTV. For the borrower,
higher LTV means lower cost in the present but higher cost in the future.
While the traditional mortgage has an LTV ratio of (at most) 80 percent, over
40 percent of subprime loans originated in 2006 had combined LTVs exceed-
ing 90 percent. LTVs were somewhat lower in the Alt-A market.38
2. Escalating Payments
The traditional FRM features a constant payment stream throughout the
loan period. In contrast, the typical subprime and Alt-A loans stipulated
monthly payments that increase over the loan period. In 2006, only 19.9
percent of first-lien subprime loans were FRMs.39 The vast majority of
loans were ARMs or hybrid mortgages that featured an initial fixed-rate
period followed by an adjustable-rate period. According to the FRB,
approximately three-fourths of originations in securitized subprime
“pools” from 2003 to 2007 were ARMs or hybrids with two- or three-
year “teaser” rates, followed by substantial increases in the rate and pay-
ment (so-called “2–28” and “3–27” mortgages).40 In 2006, the average
initial rate was 8.4 percent, while the average long-term rate, calculated
as the sum of the relevant index (usually the 6-month LIBOR) and the
contractually specified margin, was 11.4 percent.41 The expected increase
in the monthly payment at the end of the low-rate introductory period
38. For subprime LTVs—see Ben S. Bernanke, Chairman, Bd. of Governors of the Fed. Reserve
Sys., Speech at the Independent Community Bankers of America Annual Convention,
Orlando, Florida: “Reducing Preventable Mortgage Foreclosures” (March 4, 2008), available
at <http://www.federalreserve.gov/newsevents/speech/bernanke20080304a.htm> (herein-
after “Bernanke March 2008 Speech”) (basing this figure on information about loans in
securitized pools from First American LoanPerformance). For Alt-A LTVs—see Mayer et al.,
above note 23, at 33 tbl. 2B. The relevant measure is the combined LTV, which includes both
the first- and second-lien mortgages.The first-lien mortgage often has an LTV of 80 percent,
but the borrower then takes a second-lien mortgage—a piggyback loan—that further
increases the combined LTV.
39. See Demyanyk and Van Hemert, above note 1, at 1854 (tbl. 1) (counting only non-I/O, non-
balloon FRMs); see also Pennington-Cross and Ho, above note 32, at 1 (finding that, between
2003 and 2005, “the ARM market share for securitized subprime loans has ranged from just
approximately 60 percent to over 80 percent”).
40. See Truth in Lending, 73 Fed. Reg. 44,522, 44,540 (July 30, 2008) (codified at 12 C.F.R. pt. 226).
41. See Demyanyk and Van Hemert, above note 1, at 1854 (tbl. 1) (reporting the average initial
rate, 8.4 percent, and the average margin, 6.1 percent). The average long-term rate is the sum
of the margin and the index. The average value of the most popular index, the 6-month
LIBOR, was 5.3 percent in 2006. See ARM Index Values—2006 Fannie Mae LIBOR,
<http://www.efanniemae.com/sf/refmaterials/libor/index.jsp>; see also Mayer et al., above
note 23, at 11 (in 2006 and early 2007, the fully indexed rate was closer to 300 basis points
above the initial rate).
was substantial.42 Monthly payments escalated even more steeply in Alt-A

(and some prime) mortgages, where teaser rates were set further below
the market rate. These contracts stipulated an increase of up to 100 per-
cent, or $1,500 on average, in the monthly payment at the end of the
introductory period.43 According to one estimate, rate resets have
increased borrowers’ annual mortgage payments by about $42 billion.44
The escalating-payments feature was most pronounced in interest-only
(I/O) mortgages and payment-option (or simply, option) mortgages. Under
an I/O mortgage, the borrower pays interest only during the introductory
period, generally one to ten years, paying the principal only after the intro-
ductory period ends. The most popular I/O mortgages were hybrid loans,
where the introductory interest rate is fixed and the post-introductory
interest rate is variable. In 2006, approximately 20 percent of subprime orig-
inations and over 40 percent of Alt-A originations were I/O mortgages.45
An even more extreme escalating-payments contract is the option ARM.
As described by the FTC,
[o]ption ARMs generally offer borrowers four choices about how much they
will pay each month during the loan’s introductory period. Borrowers may
pay: (1) a minimum payment amount that is smaller than the amount of inter-
est accruing on the principal; (2) the amount of interest accruing on the loan
42. The actual payment shock experienced on 2005 and 2006 2–28 mortgages turned out to be
less severe, thanks to relatively low market interest rates and correspondingly low index values
in 2007 and 2008, when the interest rates on these loans reset. Still, the average monthly pay-
ment increased by more than 10 percent at reset. See Bernanke March 2008 Speech, above
note 38 (stating that even with the currently low LIBOR, a typical reset would raise the
monthly payment by more than 10 percent); Paul Willen, “Would More Disclosure of Loan
Terms Have Helped?” (presentation at FTC Mortgage Conference, May 29, 2008) 10,
available at <http://www.ftc.gov/be/workshops/mortgage/presentations/willen_paul.pdf>
(finding that payment shock for a typical subprime borrower in 2007 was 15 percent). In any
event, contractual design is determined by the ex ante expected payment shock at origina-
tion, not by the ex post actual payment shock realized two years later.
43. For Alt-A reset figures—see Cagan, above note 20, at 13 tbl. 4, 44 (showing “red” nonsub-
prime loans with steeper resets than the “orange” subprime loans; estimating a 97 percent, or
$1500, increase in the monthly payment).
44. See Cagan, id. The $42 billion figure covers the entire residential mortgage market, not only
the subprime and Alt-A segments, but ARMs and resets were common mainly in these two
segments.
45. See FTC Comment, above note 31, at 6–7 (describing I/O loans and noting the popularity
of hybrid-rate I/O loans). On the market shares of I/O loans—see Credit Suisse Report,
above note 21, at 28 (showing that I/O loans constituted $171 billion of the $824 billion in
subprime loans); see also Mayer et al., above note 23, at 7 (“Forty percent of Alt-A mortgages
involved only interest payments without any scheduled principal repayment (only about 10
percent of subprime mortgages have such an interest-only feature).”).
mortgage s 139
principal; (3) the amount of principal and interest due to fully amortize the
loan on a 15-year payment schedule; or (4) the amount of principal and inter-
est due to fully amortize the loan on a 30-year payment schedule.
Option ARMs vary in the length of the introductory periods they offer.
Some, especially in the subprime market, have introductory periods of only
one year, six months, or even one month. When the loan’s introductory
term expires, the loan is recast, amortizing to repay principal and the vari-
able interest rate over the remaining term of the loan.46
While I/O mortgages are zero-amortization loans, option ARMs imply
negative amortization by allowing below-interest monthly payments.
Accordingly, at the end of the introductory period, or even earlier, a bor-
rower might end up owing more than the value of the home. This might
happen even when home prices are steady or rising, but, of course, it is
more likely to happen when home prices are falling.47 Option ARMs were
rare in the subprime market but quite popular in the Alt-A market. By 2006
and 2007, more than 25 percent of Alt-A loans were option ARMs.48
Overall, in the Alt-A market in 2006, a large majority of originations
were nontraditional mortgage products, allowing borrowers to defer prin-
cipal or both principal and interest. These deferrals led to substantial
increases—exceeding 100 percent in some cases—in the monthly payment
at the end of the introductory period.49
3. Prepayment Penalties
Another deferred-cost component, common in subprime and Alt-A con-
tracts, is the prepayment penalty—a penalty imposed on a borrower who
repays the loan before the maturity date. Approximately 70 percent of sub-
prime loans and 40 percent of Alt-A loans included a prepayment penalty.
46. FTC Comment, above note 31, at 7.

47. See Cagan, above note 20, at 56 tbl. 30 (finding that, as of December 2006, 22.4 percent of
subprime ARMs originated between 2004 and 2006 had zero or negative equity). Another 5
percent drop in house prices, as happened after December 2006, increases the 22.4 percent
figure to 36 percent.
48. See Mayer et al., above note 23, at 13–14; see also Credit Suisse Report, above note 21, at 26,
28 (finding, based on nonagency MBS data, that in 2006, option ARMs comprised approxi-
mately 0.5 percent of the subprime market and 30 percent of the Alt-A market).
49. See Truth in Lending, 73 Fed. Reg. 44,522, 44,541 (July 30, 2008) (codified at 12 C.F.R. pt.
226) (stating that, according to one estimate, 78 percent of Alt-A originations in 2006 were
either I/O or option mortgages); GAO AMP Report, above note 26, at 14 (describing an
example with a 128 percent increase in the monthly payment at the end of the 5-year pay-
ment option period); FTC Comment, above note 31, at 9 (referring to “payment shock”).
The penalty amount is usually expressed as a percentage of the outstanding

balance on the loan, up to 5 percent, or as the sum of a specified number of
monthly interest payments, usually six.This adds up to a significant amount.
For example, a 3-percent penalty on a $200,000 balance amounts to $6,000.50
The economic importance of prepayment penalties to lenders is undeniable.
They generate substantial revenues. For example, Countrywide’s revenues
from prepayment penalties amounted to $268 million in 2006.51
Prepayment penalties can be viewed as a necessary supplement to the
escalating-payments feature: If borrowers prepay before the end of the low-
rate introductory period and thus avoid the high post-reset rates, then the
escalating-payments feature becomes moot. Prepayment penalties make it
more difficult for borrowers to avoid the escalating payments. It stands to
reason, then, that prepayment penalties played a supporting role in some
escalating-payments contracts. But in many other escalating-payments con-
tracts, this prepayment-deterrence role was less pronounced. Prepayment
penalties are generally limited to certain time periods; in other words, the
prepaying borrower will pay a penalty only if the prepayment is made dur-
ing a specified period. In many loan contracts, signed during the subprime
expansion, the prepayment-penalty period expired before the end of the
low-rate introductory period.52
Prepayment penalties are also an independent deferred-cost component of
a mortgage, regardless of their role in supporting the escalating-payments
50. On the prevalence of prepayment penalties—see Mayer et al., above note 23, at 7; see also
Demyanyk and Van Hemert, above note 1, at 1854 (tbl. 1) (showing that in 2006, 71 percent
of first-lien subprime loans included a prepayment penalty). Prepayment penalties are most
common in hybrid loans: 70 percent of hybrids have prepayment penalties, as compared to
FRMs, only 40 percent of which have prepayment penalties. See Pennington-Cross and Ho,
above note 32, at 11–12. On the size of prepayment penalties—see Michael D. Larson, “Mort-
gage Lenders Want a Commitment—and They’re Willing to Pay You for It,” Bankrate.com,
August 26, 1999, <http://www.bankrate.com/brm/news/mtg/19990826.asp> (describing
one contractual design that specifies a penalty of 3 percent of the outstanding balance for
prepayment in the first year, a 2 percent penalty for prepayment in the second year, and a 1
percent penalty for prepayment in the third year).
51. Gretchen Morgenson, “Inside the Countrywide Lending Spree”, New York Times, August 26,
2007, § 3, at 1; see also Eric Stein, Coal. for Responsible Lending, “Quantifying the Economic
Costs of Predatory Lending” (2001) 7–9, available at <http://www.responsiblelending.org/
pdfs/Quant10-01.pdf> (estimating prepayment penalty revenues at $2.3 billion each year).
52. See Michael LaCour-Little and Cynthia Holmes, “Prepayment Penalties in Residential
Mortgage Contracts: A Cost-Benefit Analysis”, Housing Pol’y Debate, 19 (2008), 631, 635 (pre-
payment penalties are generally limited in time); Mayer et al., above note 23, at 12 (“[P]repay-
ment penalties were scheduled to be in effect after the end of the teaser period for only 7
percent of the subprime short-term hybrids originated from 2003 to 2007, and over these
years the share originated with such a provision dropped from 10 to 2 percent.”).
mortgage s 141
feature.To the extent that it fails to deter prepayment, the prepayment penalty
becomes a significant cost that is deferred until the time of prepayment. This
long-term cost is linked to a reduction in the short-term cost of the loan.
Loans with prepayment penalties have lower interest rates and thus lower
monthly payments.53 Prepayment penalties thus produce the temporal-shift
characteristic of deferred-cost contracts; pay less now, pay more later.
The use of prepayment penalties has been substantially curtailed by the
Dodd–Frank Wall Street Reform and Consumer Protection Act of 2010.
The Act restricts prepayment penalties to a subset of prime, fixed-rate mort-
gages. Moreover, when allowed, prepayment penalties are restricted in
amount and duration:The prepayment penalty may not exceed 3 percent of
the prepaid amount during the first year after consummation, 2 percent
during the second year after consummation, and 1 percent during the third
year after consummation.There can be no prepayment penalty after the end
of the third year after consummation.54
B. Complexity
In addition to a variety of features that defer costs, subprime and Alt-A
mortgages during the mortgage boom were also characterized by a high
level of complexity. The complexity was the product of a proliferation of
fees and other price dimensions combined with elaborate rules governing
the application of these multiple prices.55 Beyond multidimensional pricing,
the prepayment option and the (implied) default option increased the com-
53. See Gregory Elliehausen, Michael E. Staten, and Jevgenijs Steinbuks, “The Effect of Prepay-
ment Penalties on the Pricing of Subprime Mortgages”, J. Econ. and Bus., 60 (2008), 33, 34;
LaCour-Little and Holmes, above note 52, at 642; Christopher Mayer, Tomasz Piskorski, and
Alexei Tchistyi, “The Inefficiency of Refinancing: Why Prepayment Penalties Are Good for
Risky Borrowers” (2011) Columbia Business School Working Paper. But see Engel and McCoy,
above note 44, at 2060. (Other studies found a positive, not negative, correlation between
prepayment penalty and ex ante interest rates.)
54. Pub. L. 111–203, Sec. 1414.
55. See U.S. Gen. Accounting Office, Report to the Chairman and Ranking Minority Member,
Special Committee on Aging, U.S. Senate, “Consumer Protection: Federal and State Agencies
Face Challenges in Combating Predatory Lending” Gao-04-280 (2004) 6, 21 (emphasizing
“the complexity of mortgage transactions” and the “greater variety and complexity of risks”
associated with subprime loans as compared to prime loans); Renuart and Thompson, above
note 19, at 196 (“The lender-created complexity of mortgage loans now exceeds what most
consumers, even highly educated consumers, are capable of comprehending.”); Todd J. Zywicki
and Joseph D. Adamson, “The Law and Economics of Subprime Lending” U. Colo. L. Rev. 80
(2009), 1, 55–6 (explaining that subprime loans are more complex than prime loans, and that
it is more likely that a subprime borrower will misunderstand her loan terms).
plexity involved in valuing these mortgage products. In addition, since

complexity should be measured at the market level—not at the contract
level—the existence of numerous complex products increased—exponen-
tially—the complexity of the decision that a borrower faced.
1. Interest Rates
The traditional FRM has a single interest rate that implies a constant
monthly payment. The typical subprime mortgage during the subprime
boom, the 2–28 hybrid, had an initial rate that applied for the first two years
of the loan.After the two-year introductory period expired, the loan became
an ARM with an interest rate calculated as the sum of a specified index and
a preset margin—a calculation that is repeated at the end of each adjustment
period. To make things even more complex, the loan contract commonly
specified caps that could limit the magnitude of both the periodic and total
rate adjustment.56
Other products were even more complex. As detailed above, option
ARMs commonly specified four different options for each monthly pay-
ment.These payment options were not predetermined sums; nontrivial cal-
culations were necessary to figure out what the options were. Moreover,
these contracts, while allowing negative amortization, typically capped the
level of permissible negative amortization, recasting the loan—even before
the end of the introductory period—if this cap was reached.
2. Fees
Beyond the multiple interest rates, the typical subprime and Alt-A loan
boasted a long list of fees. These fees can be divided into two categories:
origination fees and post-origination fees. Origination fees are paid at clos-
ing—that is, at the consummation of the credit transaction. Before closing a
loan contract, the lender performs credit checks and obtains appraisals to
assess the risk that it is about to undertake. The lender also commissions
various inspections, examinations, and certifications, such as pest inspection,
title examination, flood certification, and tax certification regarding the
borrower’s outstanding tax obligations.57 During the subprime boom, many
56. See Joe Peek, “A Call to ARMs: Adjustable Rate Mortgages in the 1980s”, New Eng. Econ.
Rev., March–April, (1990), 53.
57. See Elizabeth Renuart, “An Overview of the Predatory Mortgage Lending Process”, Housing
Pol’y Debate, 15 (2004), 467, 493.
mortgage s 143
lenders charged the borrower separate fees for each of these information-
acquisition services. For example, LandSafe, Countrywide’s closing-services
subsidiary, charged a $36 fee for the credit check, a $26 fee for flood certi-
fication, and a $60 fee for the tax certification. In 2006, Countrywide’s
appraisal fee revenues totaled $137 million, and its credit report fee revenues
totaled $74 million.58
Separate fees were also charged for analyzing the acquired information.
These included escrow analysis fees to cover the cost of determining the
appropriate balance for the escrow account and the borrower’s monthly
escrow payments, and underwriting analysis fees to cover the costs of ana-
lyzing a borrower’s creditworthiness. Still more fees were charged in the
forms of premiums for credit insurance, title insurance, and private mort-
gage insurance (PMI).59
Also at the closing for many of these subprime loans, the lender charged
fees for administrative services associated with the loan-origination process,
such as preparing documents, notarizing documents, and sending e-mails,
faxes, and courier mail. For example, some Countrywide loans included fees
of $45 to ship documents overnight and $100 to e-mail documents. And
then there were the general fees for such things as loan origination, loan
processing, signing documents, and closing.60 Some subprime lenders
charged up to fifteen different origination fees, which added up to thou-
sands of dollars or as much as 20 percent of the loan amount.61 The fees
were often financed into the loan amount and formed the basis for add-
itional interest charges.
58. Morgenson, above note 51.
59. See Renuart, above note 57, at 493; Willis, above note 29, at 725. According to one—now
dated—estimate, financed credit insurance costs borrowers $2.1 billion each year. See Stein,
above note 51, at 5–7.
60. Renuart, above note 57, at 493; Morgenson, above note 51.
61. See Willis, above note 29, at 786; see also HUD-Treasury Report, above note 19, at 21 (noting
origination fees of up to 10 percent of the loan amount, “far exceed[ing] what would be
expected or justified based on economic grounds”). According to HUD, borrowers are paying
excess fees averaging $700 per mortgage. See News Release, U.S. Dep’t of Hous. and Urban
Dev., “HUD Proposes Mortgage Reform to Help Consumers Better Understand Their Loan,
Shop for Lower Costs” (March 14, 2008), available at <http://www.hud.gov/news/release
.cfm?content=pr08-033.cfm.> According to Michael Kratzer, founder of FeeDisclosure.com,
a website intended to help consumers reduce fees on mortgages, of the estimated $50 billion
in transaction fees paid by mortgage borrowers (not only in the subprime and Alt-A markets),
$17 billion consist of junk fees, like $100 e-mail charges, $75 document preparation fees, and
$25 FedEx charges. See Gretchen Morgenson, “Clicking the Way to Mortgage Savings,” New
York Times, December 23, 2007, § 3, at 1. Kratzer estimates that “junk fees” have risen 50 per-
cent in recent years. See Gretchen Morgenson, “Given a Shovel, Digging Deeper into Debt,”
New York Times, July 20, 2008, at A1.
In addition to the multiple fees charged at closing, the loan contract

specified a series of future, contingent fees, including late fees, foreclosure
fees, prepayment penalties, and dispute-resolution or arbitration fees. Again,
these fees were substantial. Prepayment penalties and foreclosure fees
amounted to thousands of dollars. Late fees could amount to 5 percent of
the monthly payment.62
3. Prepayment and Default

Most mortgage contracts in the United States allow the borrower to prepay
the loan before it matures. The prepayment option may seem straightfor-
ward, but it actually adds a substantial dose of complexity to the mortgage
contract. To accurately value the contract, the borrower must estimate the
likelihood and timing of prepayment, which depend on a host of future
market conditions and personal circumstances—not to mention any exist-
ing prepayment penalty. Calculating the optimal timing for prepayment is
an example of how complex the decision can become. A common rule of
thumb for borrowers is to prepay when the expected savings from refinanc-
ing to a lower-interest loan exceeds the transaction costs associated with
terminating one loan and originating another (including the prepayment
penalty). But this rule of thumb turns out to be a very poor approximation
of the optimal prepayment decision. The reason is that the rule ignores the
option value of rejecting the current refinancing offer, even when expected
benefits exceed transaction costs, and waiting for even better refinancing
opportunities in the future.
Accounting for this option value complicates the optimal prepayment
decision. In fact, the optimal prepayment problem is so complex that it can
be solved only by high-powered computers implementing sophisticated
numeric algorithms. A closed-form approximation exists, but it, too, is far
from simple.63 In addition to the explicit prepayment option, every mort-
gage contract includes an implicit default option. The borrower can always
walk away from the mortgage, albeit at a price—that will usually include
62. See Willis, above note 29, at 725. On the magnitude of prepayment penalties—see above Part
II.A.3. On the magnitude of late fees—see Freddie Mac, Glossary of Finance and Economic
Terms, <http://www.freddiemac.com/smm/g_m.htm#L>; see also Morgenson, above note
51 (noting that, in 2006, Countrywide’s revenues from late charges amounted to $285
million).
63. See Sumit Agarwal, John C. Driscoll, and David Laibson, “Optimal Mortgage Refinancing:
A Closed Form Solution” 5–6 (2007) (Nat’l Bureau of Econ. Research,Working Paper No. 13487),
available at <http://www.nber.org/papers/w13487>.
mortgage s 145
lost equity, a damaged credit rating, and the risk of losing other assets (if the
loan is not a no-recourse loan). As with the prepayment option, valuing the
default option is a complex task.

The typical subprime or Alt-A contract, during the mortgage boom, was
multidimensional and complex. Complexity, however, should not be evalu-
ated at the single-contract level. From a functional perspective, it is more
informative to evaluate the complexity of the decision that a borrower
faces.
Borrowers must choose from among numerous mortgage products. To
make an informed choice, a borrower must read and understand numerous
complex contracts. This process would be challenging even if the compet-
ing contracts shared the same dimensions and varied only with respect to
the values assigned to each dimension. But in the subprime and Alt-A mar-
kets, the borrower had to compare different complex contracts, each with
its own set of multidimensional prices and its own rules for determining
when the different prices apply. Consider a borrower facing a 2–28 hybrid
and an option ARM: The 2–28 has an introductory period and an initial
rate. The option ARM has a different introductory period, during which
four different payment options are available.The 2–28 specifies an index and
a margin for the post-introductory period with certain caps on rate adjust-
ments.The option ARM specifies a different index, margin, and adjustment
caps. The complexity of this choice is evident. In reality, the borrower had
to choose between more than two products.64
C. Summary
In this Part, we examined several common contractual design features of
subprime and Alt-A mortgages. It should be noted that these design
features were not an innovation of the subprime expansion. For exam-
ple, relatively complex ARMs with a deferred-cost structure, created by
lower initial rates and higher long-term rates, have been offered in the
64. See William C. Apgar, Jr. and Christopher E. Herbert, U.S. Dep’t of Hous. and Urban Dev.,
“Subprime Lending and Alternative Financial Service Providers: A Literature Review and
Empirical Analysis” § 2.2.3 (2006) (describing “the bewildering array of mortgage products
available”).
prime market since the early 1980s.65 While cost deferral and high levels of
complexity are not unique to subprime loans, these design features have
been enhanced in subprime and Alt-A contracts. Since complex deferred-
cost loans have been around for a while, they cannot be the only cause of
the subprime expansion and the ensuing subprime crisis. They are probably
not even the main cause. But, as we’ll see in the following Parts, these com-
plex, deferred-cost loans did play an important role in the rise and fall of the
subprime market.
III. Rational-Choice Theories and Their Limits

Why were subprime mortgage contracts designed to defer costs? Why was
the total cost of the loan divided into so many different interest rates and
fees? In this Part, we’ll evaluate the standard rational-choice explanations for
these contractual design features.The rational-choice theories explain some
of the observed practices in the subprime market, but there is much that
they cannot explain. This explanatory gap will be filled in Part IV by a
behavioral-economics theory.
A. Deferred Costs
1. Affordability
Perhaps the most common justification for deferred-cost contracts is afford-
ability. If borrowers cannot afford to make a substantial down payment, then
they will take a mortgage with a high LTV. If they currently cannot afford
to make high monthly payments, then they will take a mortgage with low
initial monthly payments. Deferred-cost contracts create short-term afford-
ability.66 Indeed, by most accounts, deferred-cost contracts were designed to
65. See Peek, above note 56, at 50, 54; see also Zywicki and Adamson, above note 55, at 5–7 (explain-
ing how legal reform in the early 1980s—specifically the Alternative Mortgage Transaction
Parity Act of 1982—lifted severe restrictions on the design of mortgage contracts). Moreover,
deferred-cost loans are common in other countries (interest-only mortgages are standard in the
United Kingdom) and in other sectors (corporate bonds are designed as interest-only loans).
66. See GAO AMP Report, above note 26, at Abstract (“Federally and state-regulated banks and
independent mortgage lenders and brokers market AMPs [mostly I/O and payment-option
loans], which have been used for years as a financial management tool by wealthy and finan-
cially sophisticated borrowers. In recent years, however, AMPs have been marketed as an
mortgage s 147
secure short-term affordability. But short-term affordability is not a rational-

choice explanation. For affordability to be a rational-choice explanation for
cost deferral, long-term affordability must be the key consideration. In other
words, the borrower must be able to service the loan both now and in the
future.While deferred-cost contracts clearly enhance short-term affordabil-
ity, it is by no means clear that they enhance long-term affordability. Paying
less now means paying more later. Smaller down payments (higher LTVs)
and lower initial payments are followed by higher monthly payments in the
future. Affordability in the long term can rationally explain deferred-cost
contracts only if the borrower’s available income is expected to increase as
fast as (or faster than) the escalating mortgage payments.67
In this spirit, the FRB advises borrowers that “[d]espite the risks of these
loans, an I-O mortgage payment or a payment-option ARM might be right
for you if . . . you have modest current income but are reasonably certain
that your income will go up in the future (for example, if you’re finishing
your degree or training program). . . . ”68 But how many borrowers fit this
description? Notice that the FRB is not talking about standard, gradual pay
raises.Those would not match the substantial increase in the monthly mort-
gage payment at the end of the introductory period that many subprime
and Alt-A contracts stipulated. The FRB is referring to students and train-
‘affordability’ product to allow borrowers to purchase homes they otherwise might not be
able to afford with a conventional fixed-rate mortgage.”); Mayer et al., above note 23, at 7
(“[S]ubprime borrowers may have turned to these products in an attempt to obtain more
affordable monthly payments.”). Affordability concerns were especially acute in areas where
rapidly rising home prices forced borrowers to take larger loans.
67. The failure to adopt this long-term affordability perspective has been the subject of criticism.
In particular, lenders have been criticized for qualifying borrowers who can make the low
short-term payments but not the high long-term payments. See Hearing, above note 1, at 11
(“Some subprime lenders . . . established borrowers qualification for mortgages on the basis of
initially low teaser rates.”).This concern was addressed by the ability-to-repay requirement in
Sec. 1402 of the Dodd—Frank Act. See also Truth in Lending, 73 Fed. Reg. 44,522, 44,539
(July 30, 2008) (codified at 12 C.F.R. pt. 226). The effect of these regulations, had they come
sooner, could have been substantial. In a presentation to investors, Countrywide Financial
acknowledged that it would have refused 89 percent of its 2006 borrowers and 83 percent of
its 2005 borrowers, representing $138 billion in mortgage loans, had it followed the long-term
affordability standards adopted in the FRB’s regulations. See Binyamin Appelbaum and Ellen
Nakashima, “Banking Regulator Played Advocate over Enforcer: Agency Let Lenders Grow
out of Control, Then Fail,” Wash. Post, November 23, 2008, at A1.
68. See Bd. of Governors of the Fed. Reserve Sys. et al., “Interest-Only Mortgage Payments and
Payment-Option ARMs—Are They for You?” (2006) 7, available at <http://www.federalreserve
.gov/pubs/mortgage_interestonly/mortgage_interestonly.pdf> (hereinafter FRB, Interest
Only); see also FTC Comment, above note 31, at 8 (noting the advantage of alternative mort-
gage products for “upwardly mobile” borrowers).
ees. Indeed, 2–28 hybrids, and even I/O and option mortgages, may be bene-
ficial for a second-year law student who anticipates a sharp increase in
income after graduation.These students and trainees are good candidates for
escalating-payments contracts, yet there are too few of them to explain a
significant fraction of the approximately two million hybrid loans origin-
ated per year at the height of the subprime market.69
While borrowers with rising incomes are the natural candidates for esca-
lating-payments contracts, borrowers with variable incomes may also find
some of these contractual designs beneficial. The FRB advises that a bor-
rower with volatile income who can afford to make only small monthly
payments in low-income periods may rationally prefer a loan contract that
requires lower monthly payments.70 But the typical loan does not offer the
low-payment option for more than two years. Accordingly, the income of
the target borrower should be volatile only temporarily. Moreover, rational
borrowers with volatile income should have no problem making fixed-
amount mortgage payments—all they need to do is save some of their earn-
ings from the high-income periods. As with rising-income borrowers, the
number of variable-income borrowers who would benefit from deferred-
cost loans seems small relative to the number of loans with these design
features.
The long-term affordability explanation covers a small fraction of
deferred-cost originations. This assessment is consistent with the evidence
of especially high foreclosure rates on homes financed by deferred-cost
loans.71 If deferred-cost loans were designed to address short-term liquidity
problems, then defaults and foreclosures would be rare. But perhaps there is
another, more plausible version of the affordability explanation. Thus far,
long-term affordability has been assumed to imply an ability to make the
high future payments from rising income. A less literal interpretation of
affordability may include an expectation to avoid, rather than actually make,
69. The two million estimate is based on the 2,780,000 first-lien subprime loans originated in
2006, see above Part I.B, multiplied by the 75 percent of hybrid ARMs among subprime
loans. See above Part II.A.2.
70. See FRB, Interest Only, above note 68 (advising borrowers that I/O loans and option ARMs
may be suitable for them if they “have irregular income (such as commissions or seasonal
earnings) and want the flexibility of making I-O or option-ARM minimum payments dur-
ing low-income periods and larger payments during higher-income periods”); see also FTC
Comment, above note 31, at 8 (noting the advantage of alternative mortgage products for
borrowers with variable income).
71. See Paulson, above note 25.
mortgage s 149
the high future payments by refinancing the loan before the low-rate intro-
ductory period ends.72
There are three circumstances under which a borrower could expect to
obtain a new mortgage with lower monthly payments:
(1) The borrower’s credit score improves by regularly making the low
payments during the introductory period.73
(2) The market interest rate falls.
(3) House prices increase, resulting in a lower LTV for the new mortgage.
The question then is how many borrowers rationally expected that such
positive developments would enable them to refinance their deferred-cost
mortgage and avoid the high long-term costs. From an ex post perspective,
it is clear that the subprime crisis and the ensuing tightening of credit elimin-
ated the refinancing option for many borrowers. The FRB infers that even
from an ex ante perspective, which is the relevant perspective for judging
the affordability explanation, many borrowers could not have rationally
expected to have the opportunity to benefit from attractive refinancing
options:
[E]vidence from recent events is consistent with a conclusion that a wide-
spread practice of making subprime loans with built-in payment shock after
a relatively short period on the basis of assuming consumers will accumulate
sufficient equity and improve their credit scores enough to refinance before
the shock sets in can cause consumers more injury than benefit.74
72. See Truth in Lending, 73 Fed. Reg. 1672, 1687 (January 9, 2008) (codified at 12 C.F.R. pt. 226)
(“Consumers may also benefit from loans with payments that could increase after an initial
period of reduced payments if they have a realistic chance of refinancing, before the payment
burden increases substantially, into lower-rate loans that were more affordable on a longer-
term basis. This benefit is, however, quite uncertain, and it is accompanied by substantial
risk....”); FTC Comment, above note 31, at 8 (“[B]orrowers who are confident they will sell
or refinance their homes for an equal or increased value before the introductory period of the
loan expires may benefit from alternative loan options.”). Prepayment to avoid high post-reset
rates was common before the subprime crisis hit and the credit crunch set in. See Penning-
ton-Cross and Ho, above note 32, at 10 (finding, based on LP data, that hybrid mortgages tend
to prepay quickly around the first mortgage reset date); Shane M. Sherlund,“The Past, Present,
and Future of Subprime Mortgages” (2008) 10 (Bd. of Governors of the Fed. Reserve Sys. Fin. and
Econ. Discussion Series Paper No. 2008–63) (finding that “prepayments jump during reset
periods”).
73. See Mayer et al., above note 23, at 11 (“Industry participants claim that teaser mortgages were
never designed as long-term mortgage products. Instead, they argue that the two- or three-
year teaser period was designed for consumers with tarnished credit to improve their credit
scores by making regular payments....”).
74. Truth in Lending, 73 Fed. Reg. at 1688.
The possibility of refinancing and prepayment, together with short-term

affordability concerns, can also explain the prevalence of another
deferred-price dimension: prepayment penalties. The prepayment option
benefits borrowers. But borrowers must pay for this benefit. One way
they pay is through a higher initial interest rate. Short-term affordability
concerns render this ex ante payment unattractive. The alternative is to
pay for the prepayment option ex post with a prepayment penalty. In
other words, the penalty reduces the value of the prepayment option to
the borrower but also reduces the cost that this option imposes on the
lender. This explains the lower interest rates on loans with prepayment
penalties.75 While this explanation for the prevalence of prepayment pen-
alties is persuasive, it is incomplete, because it implies that prepayment
penalties replace higher interest rates. There is evidence, however, that
the amounts paid in penalties ex post exceed the foregone interest pay-
ments that were not paid ex ante.76
2. Speculation
An alternative rational-choice explanation portrays the deferred-cost mort-
gage as an investment vehicle designed to facilitate speculation on real-estate
prices.77 This explanation applies to the substantial portion—10 percent in
the subprime market and 25 percent in the Alt-A market—of loans that were
originated on investment properties.78 It may also apply to loans originated
on owner-occupied properties. The speculator purchases a house with a
deferred-cost mortgage and begins making the initial, low monthly pay-
ments. If real-estate prices go up, the speculator will either sell the house and
pocket the difference between the lower buy price and the higher sell price
or refinance the loan using the increased equity to obtain lower long-term
rates. If real-estate prices go down, the speculator will simply default on the
75. See above Part II.A.3.

76. See LaCour-Little and Holmes, above note 52, at 662 (comparing 2–28 ARMs with lower
initial rates and prepayment penalties to 2–28 ARMs with higher initial rates and without
prepayment penalties, and finding that the total interest-rate savings is significantly less than
the amount of the expected prepayment penalty). And according to some studies there is even
no negative correlation between prepayment penalties and ex ante interest rates. See above
Part II.A.3.
77. I focus on the effects of home-price trends and expectations about home-price trends. A
similar argument can be made about market interest rates and expectations about market
interest rates.
78. Mayer et al., above note 23, at 19 (reporting the shares of loans originated on investment
properties in the subprime and Alt-A markets).
mortgage s 151
mortgage. The speculator enjoys the upside benefit, while the lender bears
the downside cost. This attractive arrangement is purchased at the bargain
price of the low, initial payments on a deferred-cost mortgage. The high,
long-term costs are avoided.79
Speculation with the help of deferred-cost loans, however, is not really a
risk-free prospect. The speculator does not simply default on the mortgage.
Default is costly. First, in jurisdictions where the lender has recourse to the
borrower’s assets, default places these assets at risk. It is important to note,
however, that a large number of states, including subprime hot spots like
California, Colorado, Nevada, and Arizona, have no-recourse laws. And
even in states without no-recourse laws, filing an action for deficiency is
often not cost-effective for the lender, and thus the loan becomes a de facto
no-recourse loan.
A second cost of default is foregone equity, although this cost is also often
small due to high initial LTVs and even higher LTVs at the time of
default (recall that default is triggered by falling house prices). A third cost
of default is the damage to the borrower’s credit rating and the increased
future cost of credit that a damaged credit rating implies. Finally, default
entails foreclosure and relocation—both costly prospects. While there is no
consensus estimate for the cost of default and foreclosure, for many borrow-
ers this cost will amount to tens of thousands of dollars. These costs may
explain why strategic default (when house prices fall below the outstanding
loan amount) is relatively rare.80
Despite the cost of default, however, the downside risk is still outweighed
by the upside benefit as long as the probability of a positive result is suffi-
ciently high. In other words, if house prices are expected to rise high enough
79. Adopting the “heads—borrower wins, tails—lender loses” strategy is rational for borrowers
but not for lenders. The speculation explanation is incomplete absent an account of lenders’
incentives. Why did lenders play along? Agency problems—within lending institutions and
among the different parties in the securitization process—provide one set of answers. See
above note 10 and accompanying text.
80. On the issue of recourse—see Michael T. Madison, Jeffry R. Dwyer, and Steven W. Bender,The
Law of Real Estate Financing vol. 2 § 12:69 (Thomson Reuters/West, rev. edn. 2008); Zywicki
and Adamson, above note 55, at 29 n.134 (estimating that about fifteen to twenty states, includ-
ing many larger states, have no-recourse laws); Zywicki and Adamson, id. at 30 (on de facto
no-recourse). On the cost of a damaged credit rating—see Kenneth P. Brevoort, and Cheryl R.
Cooper, Foreclosure’s Wake: The Credit Experiences of Individuals Following Foreclosure
(October 12, 2010), available at <http://ssrn.com/abstract=1696103>. On the relatively low
frequency of strategic default—see Luigi Guiso, Paola Sapienza, and Luigi Zingales, “Moral and
Social Constraints to Strategic Default on Mortgages” (2009) NBER Working Paper No. 15145.
and fast enough, then speculation is rational even if the costs incurred in the
unlikely event of default are substantial.81
The question, therefore, is whether such borrower expectations of a con-
tinuing, rapid increase in house prices were rational. An initial observation is
that during the subprime expansion, home prices were high relative to
underlying fundamentals. As noted by Peter Orszag, the CBO director, “for
a time, the expectation of higher prices became a self-fulfilling prophecy that
bore little relation to the underlying determinants of demand, such as demo-
graphic forces, construction costs, and the growth of household income.”82
But expectations that deviate from long-term fundamentals are not necessar-
ily irrational. For example, a rational borrower may recognize that home
prices must fall eventually but expect to exit the market before the correc-
tion. This expectation, while it proved to be erroneous for many subprime
and Alt-A borrowers ex post, may well have been rational ex ante.
There were surely rational speculators in the subprime and Alt-A mar-
kets who rode the real estate bubble armed with accurate ex ante estimates
(that turned out to be false ex post) about the timing of the bubble’s inevi-
table end. But there were also other borrowers/speculators with optimistic
expectations of future house prices that were not rationally formed. Specifi-
cally, the irrational borrowers extrapolated from past price trends: If home
prices increased by 10 percent over the past year, these borrowers expected
that home prices would also increase by 10 percent over the next year.
Indeed, in an influential study, Karl Case and Robert Shiller found that
many home buyers overestimate the correlation between past trends and
future price movements. In other words, backward-looking tendencies drive
expectations of future price growth (beyond what could plausibly be justi-
fied in a rational-expectations model).83 The subprime and Alt-A markets
81. The upside benefit is also not as straightforward as implied in the initial description. Sale and
refinancing involve transaction costs and, in many cases, also prepayment penalties. Moreover,
even with increasing house prices, a borrower may be left with low or negative equity, the
result of high initial LTVs and slow—zero or even negative—amortization, severely reducing
sale and refinancing options. But, again, this only means that a rational speculator must have
expected a substantial increase in house prices—an increase sufficient to outweigh the costs
and difficulties of sale and refinancing.
82. Hearing, above note 1, at 10. See also Robert J. Shiller, “Understanding Recent Trends in
House Prices and Home Ownership” (2007) 4–5 (Yale Univ. Econ. Dep’t,Working Paper No. 28),
available at <http://www.econ.yale.edu/ddp/ddp25/ddp0028.pdf>.
83. See Karl E. Case and Robert J. Shiller, “The Behavior of Home Buyers in Boom and
Post-Boom Markets”, New. Eng. Econ. Rev., (November-December 1988) 29; see also Karl
E. Case and Robert J. Shiller, “Is There a Bubble in the Housing Market?” (2003) Brook-
ings Papers on Econ. Activity, No. 2, at 299; Robert J. Shiller, “Speculative Prices and Popular
mortgage s 153
experienced both rational and irrational speculation.84 The relative propor-

tion of these two species of speculators remains an open question.
B. Complexity
1. Interest Rates
Mortgage loans, like any other long-term credit product, are subject to
interest-rate risk—the risk that market interest rates will change over the
life of the loan, departing, often substantially, from the interest rates that
prevailed at the time of origination. In a rational-choice framework,
ARMs, with their complex formulas for setting interest rates, are designed
to optimally allocate interest-rate risk between the lender and the bor-
rower. While an FRM allocates all interest-rate risk to the lender, a pure
ARM, with an interest rate that closely tracks a market index, provides the
polar opposite allocation, imposing all the interest-rate risk on the bor-
rower. The more complex and common ARMs, with caps that limit inter-
est rate adjustments, enable a range of risk allocations between these two
extremes.
Models”, J. Econ. Persp., Spring (1990), at 55, 58–61. Moreover, Case and Shiller found that
many home buyers believe that home prices cannot decline. Id., at 59. Case and Shiller
repeated their study for the recent housing bubble, obtaining similar results. See Karl E.
Case and Robert J. Shiller, “Home Buyer Survey Results 1988–2006” (unpublished paper,
Yale University, 2006); see also Shiller, above note 82, at 11. These survey results are also
supported by evidence of borrower behavior. In particular, home buyers extend them-
selves more—via higher LTVs and higher payment-to-income ratios—when buying a
home in markets with high historical appreciation rates. See Christopher Mayer and
Todd Sinai, “Housing and Behavioral Finance” (2007) (unpublished manuscript), availa-
ble at <http://real.wharton.upenn.edu/~sinai/papers/Housing-Behavioral-Boston-Fed
-v9.pdf>. Mayer and Karen Pence found that “[a] one standard deviation increase in house
price appreciation in [2004] is associated with a 39 percent increase in subprime loans [in
2005].” Christopher J. Mayer and Karen Pence, “Subprime Mortgages: What, Where, and
to Whom?” (2008) (Nat’l Bureau of Econ. Research, Working Paper No. 14083), available at
<http://ssrn.com/abstract=1149330>. Since it is a lagged appreciation variable that is
correlated with the increase in subprime originations, this finding is consistent with a
behavioral story that demand for subprime loans was driven by expectations of future
house-price appreciation based on extrapolation from past trends.
84. This is consistent with a leading economic theory of bubbles, which posits the existence of both
rational and irrational traders. See J. Bradford De Long et al., “Noise Trader Risk in Financial
Markets”, J. Pol. Econ.’y 98 (1990), 703, 705; J. Bradford De Long et al.,“Positive Feedback Invest-
ment Strategies and Destabilizing Rational Speculation”, J. Fin., 45 (1990), 379, 380; Andrei
Shleifer and Lawrence H. Summers, “The Noise Trader Approach to Finance”, J. Econ. Persp.,
Spring (1990), at 19, 28–9; see also Robert J. Shiller, Irrational Exuberance (Princeton University Press,
2000) 60–4 (developing a market-psychological theory of bubbles).
ARMs were initially developed in the early 1980s to protect lenders from
the interest-rate risk that they bore under the traditional FRM.85 In a time
when loan originators held mortgages on their own balance sheets, shifting
the risk to the borrower was an important means of minimizing the risk.
This explanation for ARMs is less powerful, however, in the era of securiti-
zation. Originators no longer bear much, if any, of the interest-rate risk.The
securitizers spread this risk among multiple investors, who are usually better
situated to bear this risk than the typical borrower.
2. Fees
As explained in Part II, many different services and costs are associated with
the mortgage transaction. In the past, most of these costs were folded into
the loan’s interest rate. During the subprime expansion, lenders and their
affiliates—mortgage settlement/closing companies and servicers—charged
separate fees for each service rendered or cost incurred. There are two
rational-choice, efficiency-based explanations for the proliferation of fees.
First, to the extent that some services are optional, setting separate prices
for these services allows for more efficient tailoring of the product to the
needs and preferences of different borrowers. This explanation is plausible
for some services and fees but not for others. Specifically, it is not plausible
for the many non-optional services that all borrowers purchase, such as
credit checks, document preparation, and appraisals. Moreover, evidence of
“[w]ild variation” in fees charged for largely standardized services is incon-
sistent with a claim that borrowers paid the cost of optional services that
they requested.86
The second rational-choice explanation describes the proliferation of
fees in subprime mortgage contracts as reflecting a desirable shift to risk-
based pricing. For example, if the costs of delinquency and foreclosure
proceedings are folded into the interest rate, then non-defaulting
borrowers will pay for the delinquency and foreclosures of defaulting
borrowers. Separate late fees and foreclosure fees eliminate this cross-
subsidization. Again, this explanation is plausible for certain fees, but not
for others.
85. See Peek, above note 56, at 48.

86. See Mark D. Shroder, “The Value of the Sunshine Cure: The Efficacy of the Real Estate Settle-
ment Procedures Act Disclosure Strategy”, Cityscape: J. Pol’y Dev. and Res., No. 1, (2007) at 73,
84 (noting, for example, that the cost of obtaining a credit report, “a standard national, largely
automated, service” is typically about $50, yet credit report fees range from $25 to $100).
mortgage s 155
3. Prepayment and Default

The implied default option is an inevitable component of any loan product.
So let’s focus on the prepayment option that, while ubiquitous in mortgage
contracts in the United States, is much more limited in most other coun-
tries.87 The prepayment option serves two main goals. First, by allowing
borrowers who improve their credit ratings to refinance into lower-rate
loans, the prepayment option allows individuals to achieve home ownership
earlier. Second, the prepayment option protects borrowers from the risk of
paying a mortgage interest rate that is substantially above the current market
rate. These benefits, however, should be weighed against the difficulty of
valuing a mortgage with a prepayment option.

The decision problem faced by a potential borrower was made difficult by
the complexity of the typical subprime or Alt-A mortgage and even more
difficult by the need to choose among multiple complex mortgage prod-
ucts. The standard efficiency explanation for the large variety of products
available in many markets is consumer heterogeneity. In the mortgage mar-
ket, different borrowers have different preferences and face different con-
straints. A mortgage design that is ideal for one borrower could be terrible
for another. With more products to choose from, each borrower, in theory,
can choose the mortgage that is best for her. This explanation, however,
assumes that informed choice is possible, despite the high level of complex-
ity of the choice problem.88
C. Summary
Efficiency-based rational-choice theories can explain many, though not all,
of the contractual design features observed in the subprime and Alt-A mar-
kets. Moreover, even for the design features that can be explained within a
rational-choice framework, the rational-choice theories have limited reach.
Rational-choice theories explain the demand structure of rational borrow-
87. Richard K. Green and Susan M. Wachter, “The American Mortgage in Historical and Inter-
national Context”, J. Econ. Persp., Fall (2005) 93, 101.
88. The rational-choice account recognizes that complexity—of a single product and of the array
of offered products—increases the cost of shopping. When shopping costs more, the rational
borrower will shop less. Since shopping creates a positive externality, there is a risk that the
market will produce an inefficiently high level of complexity.
ers and the contractual-design response to this demand. As we will see in

Section IV.C., however, not all borrowers, and especially not all subprime
and Alt-A borrowers, were financially sophisticated, rational borrowers.
Rational-choice theories leave an explanatory gap. Let’s look at how to fill
this gap.
IV. A Behavioral-Economics Theory

The subprime mortgage contract is a product of the interaction between
the forces of supply and demand in the subprime mortgage market. When
lenders respond to a demand for financing that is influenced by borrower
psychology, the resulting loan contract will feature deferred costs and a high
level of complexity. The recent history of the subprime market thus serves
as an example of the general theory proposed in Chapter 1.
A. Deferred Costs
As we have seen in previous chapters, the behavioral-economics explanation
for deferred-cost contracts is based on evidence that future costs are often
underestimated. When future costs are underestimated, contracts with
deferred-cost features become more attractive to borrowers and thus to
lenders. Consider a simplified loan contract with two price dimensions: a
short-term price, PST , and a long-term price, PLT . Assume that the optimal
mortgage contract sets PST = 5 and PLT = 5, as these prices provide optimal
incentives and minimize total costs. If borrowers are rational, lenders will
offer this optimal contract.
Now assume that borrowers underestimate future costs. Assume that
borrowers perceive the long-term payments to be one-half of the actual
1
long-term payments: PˆLT = 2 ⋅ PLT ⋅ As a result of such misperception,
lenders will no longer offer the optimal contract. To see this, compare the
optimal (5,5) contract, with an inefficient, deferred-cost contract setting
PST= 3 and PLT= 8, the (3,8) contract. Assume that under both contracts, the
lender just covers the total cost of making the loan. (The total cost is higher
under the inefficient, deferred-cost contract: 8 + 3 > 5 + 5.) Total payments
under the optimal contract, as perceived by the imperfectly rational borrow-
ers, would be P̂(5, 5) = 5 + 1 ⋅ 5 = 7.5. (Time discounting is ignored for simpli-
2
mortgage s 157
city.) Perceived total payments under the inefficient, deferred-cost contract

1
would be P̂(3, 8) = 3 + 2 ⋅ 8 = 7. Borrowers would prefer, and thus lenders
will offer, the inefficient, deferred-cost contract.89
There are several reasons to expect systematic underestimation of future
costs. Myopia is one such reason. High LTV contracts are attractive to myopic
borrowers, who place excessive weight on the short-term benefits of a low
down payment (or a large cash-out in a refinance loan) and insufficient
weight on the long-term consequences of a high LTV, such as higher interest
payments and greater difficulty to refinance. Escalating-payments contracts
are similarly attractive to myopic borrowers, who are attracted to the initial
low payments while failing to consider the future high payments. Myopia
will also lead borrowers to discount the costs associated with a prepayment
penalty—either the penalty itself or the cost of delayed prepayment.
Another bias responsible for the underestimation of future costs is opti-
mism. Borrowers might be optimistic about their future income.They might
also optimistically underestimate the probability of an adverse contingency,
such as job loss, accident, or illness, causing them financial hardship. As a
result, borrowers might overestimate their ability to service a loan with
high, deferred costs. In addition, borrowers might overestimate their ability
to refinance the loan at an attractive rate and to avoid the high, long-term
costs associated with a deferred-cost loan by doing so. Such overestimation
may result from optimism about future home prices, future interest rates,
and the borrower’s future credit score.
During the subprime boom, some borrowers were both myopic and opti-
mistic. Moreover, some lenders and brokers reinforced borrowers’ myopia
89. As noted by the Federal Reserve Board when revising its Truth in Lending regulations:
A consumer may focus on loan attributes that have the most obvious and immediate con-
sequence such as loan amount, down payment, initial monthly payment, initial interest
rate, and up-front fees (though up-front fees may be more obscure when added to the loan
amount, and “discount points” in particular may be difficult for consumers to understand).
These consumers, therefore, may not focus on terms that may seem less immediately
important to them such as future increases in payment amounts or interest rates, prepay-
ment penalties, and negative amortization. . . . Consumers who do not fully understand
such terms and features, however, are less able to appreciate their risks, which can be sig-
nificant. For example, the payment may increase sharply and a prepayment penalty may
hinder the consumer from refinancing to avoid the payment increase. Thus, consumers
may unwittingly accept loans that they will have difficulty repaying.
Truth in Lending, 73 Fed. Reg. at 44,525–26. See also David Miles, “The U.K. Mortgage Market:
Taking a Longer-TermView, Interim Report: Information, Incentives and Pricing” (2003) 3, avail-
able at <http://www.hm-treasury.gov.uk/consult_miles_index.htm> (noting that borrowers tend
to focus disproportionately on the initial, rather than the long-term, cost of a loan).
and optimism.90 These biases provide an alternative, behavioral explanation

for the prevalence of cost deferral. Myopia and optimism explain why short-
term affordability, rather than rational, long-term affordability, took center
stage in the subprime and Alt-A markets. These biases—especially optimism
about future house prices—also add an important dose of reality to the
speculation explanation.
B. Complexity
The typical subprime and Alt-A mortgage contract during the boom years
was complex. It specified numerous interest rates, fees, and penalties, the
magnitude and applicability of which were often contingent on unknown
future events. Rational borrowers could navigate this complexity with ease.
They could accurately assess the probability of triggering each rate, fee, and
penalty, and then accurately calculate the expected magnitude of each rate,
fee, and penalty. Accordingly, each price dimension would have been
afforded the appropriate weight in the overall evaluation of the mortgage
product.
Imperfectly rational borrowers, however, were incapable of such accurate
assessments. They were unable to calculate prices that were not directly
specified. Even if they could have performed this calculation, it is unlikely
that they would have been able to simultaneously consider 10 or 15 (or even
more) price dimensions. And even if they could have remembered all the
price dimensions, they probably would have been unable to calculate the
impact of these prices on the total cost of the loan. While the rational
226) (“In addition, originators may sometimes encourage borrowers to be excessively opti-
mistic about their ability to refinance should they be unable to sustain repayment. For exam-
ple, they sometimes offer reassurances that interest rates will remain low and house prices will
increase; borrowers may be swayed by such reassurances because they believe the sources are
experts.”); see also Complaint, People v. Countrywide Fin. Corp., (Cal. Super. Ct. June 24, 2008)
(claiming that Countrywide encouraged borrowers to take complex hybrid and option
ARMs by emphasizing low teaser rates and misrepresenting long-term costs) (the complaint,
the California settlement, signed by the California Attorney General on October 6, 2008, and
the Multistate Settlement Term Sheet, signed by the Attorneys General of Arizona, Connecti-
cut, Florida, Illinois, Iowa, Michigan, Nevada, North Carolina, Ohio, and Texas, can be found
at <http://www.consumerlaw.org/unreported>); Gretchen Morgenson,“Countrywide Sub-
poenaed by Illinois,” New York Times, December 13, 2007, at C1 (stating that the Illinois Attor-
ney General sued a Chicago mortgage broker and is investigating Countrywide Financial, the
broker’s primary lender, for abusive lending practices, specifically pushing borrowers into
payment-option ARMs by emphasizing the low short-term payments and deemphasizing the
high long-term costs).
mortgage s 159
borrower was unfazed by complexity, the imperfectly rational borrower

might have been misled by complexity.
As we have seen, imperfectly rational borrowers often deal with com-
plexity by ignoring it.They simplify the decision problem by ignoring non-
salient price dimensions and approximating, rather than calculating, the
impact of the salient dimensions that cannot be ignored. In particular, lim-
ited attention and limited memory might result in the exclusion of certain
price dimensions from consideration. And limited processing ability might
prevent borrowers from accurately aggregating the different price compo-
nents into a single, total expected price that would serve as the basis for
choosing the optimal loan.91
Increased complexity may be attractive to lenders because it allows them
to hide the true cost of the loan in a multidimensional pricing maze. A
lender who understands the imperfectly rational response to complexity
can use complexity to its advantage—to create an appearance of a lower
total price without actually lowering the price. For example, if the tax cer-
tification fee and the late-payment fee are not salient to borrowers, lenders
will raise the magnitude of these price dimensions. Increasing these prices
will not hurt demand. On the contrary, it will enable the lender to attract
borrowers by reducing the magnitude of more salient price dimensions.
This strategy depends on the existence of non-salient price dimensions.
When the number of price dimensions goes up, the number of non-salient
price dimensions can also be expected to go up. Lenders thus have a strong
incentive to increase complexity and multidimensionality.
Lenders also have a strong incentive to increase the complexity of salient
price dimensions, like the options in an option ARM and the adjusting
interest rate in a 2–28 hybrid with adjustment caps. The borrower who is
unable to calculate these prices will try to approximate them. This makes
complexity attractive to lenders because the borrower’s approximation is
usually an underestimation.
91. See GAO AMP Report, above note 26, at Abstract (“Regulators and others are concerned
that borrowers may not be well-informed about the risks of AMPs, due to their complexity
and because promotional materials by some lenders and brokers do not provide balanced
information on AMPs benefits and risks.”); FTC Comment, above note 31, at 14 (“[F]or loans
with more complexity—such as nontraditional mortgages—consumers face further chal-
lenges in understanding all significant terms and costs.”); Hearing, above note 1, at 13 (the
CBO suggested that “[t]he rise in defaults of subprime mortgages may also reflect the fact
that some borrowers lacked a complete understanding of the complex terms of their mort-
gages and assumed mortgages that they would have trouble repaying.”).
Finally, complexity can be expected to increase as borrowers learn to

effectively incorporate more price dimensions into their decision. If lenders
significantly increase the magnitude of a non-salient price dimension, bor-
rowers will eventually learn to focus on that price dimension and, eventu-
ally, it will become salient. Lenders will have to find another non-salient
price dimension. When they run out of non-salient prices in the existing
contractual design, they may create new ones by adding more interest rates,
fees, or penalties. Similarly, borrowers will eventually learn to accurately
estimate those prices that, while salient, are indirectly defined using com-
plex formulae and whose impact depends on a host of unknown future
realizations. When this happens, lenders will have an incentive to increase
even further the complexity of these or other prices.
C. Heterogeneity in Cognitive Ability

The limits of the rational-choice theories, explored in Part III, open the
door to the consideration of an alternative, behavioral-economics theory.
By integrating psychology and economics, this theory can better explain
the contractual design features observed in the subprime and Alt-A mort-
gage markets. But the two theoretical approaches—the neoclassical, rational-
choice approach and the behavioral approach—are not mutually exclusive.
The rational-choice theories explain the behavior of the more sophisticated
borrowers and the market’s response—specifically the contractual-design
response—to the demand generated by these borrowers. Meanwhile, the
behavioral-economics theory explains the demand generated by less sophis-
ticated borrowers and how lenders designed their contracts in response to
this demand.92
The relative domain of the two competing theoretical approaches can be
indirectly assessed using evidence on the cognitive abilities of borrowers.
Available evidence suggests that imperfect rationality is pervasive in the resi-
dential mortgage market and especially in the subprime market. A recent
study, by Sumit Agarwal, Gene Amromin, Itzhak Ben-David, Souphala
Chomsisengphet, and Douglas D. Evanoff, found that mandated financial
92. Cf. Allie Schwartz, “Who Takes Out Adjustable Rate Mortgages?” (2009) Harvard University
Working Paper (2009) (cited in Daniel Bergstresser and John Beshears, “Who Selected Adjusta-
ble-Rate Mortgages?: Evidence from the 1989–2007 Surveys of Consumer Finances”, (2010)
HBS Working Paper 10-083, 5) (showing that the ARM market is split into two very different
submarkets—a high-income, wealthy segment and a low-income, credit-constrained segment).
mortgage s 161
counseling is correlated with less risky ARM contracts, specifically with

higher short-term teaser rates and lower long-term rates.93 These coun-
seling sessions likely address both an information and a cognitive deficit
among borrowers.
Survey studies and consumer testing conducted by the FRB and the
FTC found that borrowers simply do not understand mortgage terms. Also,
in testing the efficacy of proposed disclosures, the FTC identified substantial
framing effects; different disclosure forms containing the same information
led to different choices—a result that would not be expected if borrowers
were perfectly rational.94 A recent survey study, by Annamaria Lusardi, con-
cludes that many individuals are not well informed and knowledgeable
about their terms of borrowing and that a sizeable group does not know the
terms of their mortgages. Lusardi also found that “the majority of Ameri-
cans lack basic numeracy and knowledge of fundamental economic princi-
ples, such as the workings of inflation, risk diversification, and the relationship
between asset prices and interest rates.”95
93. Sumit Agarwal et al., “Can Mandated Financial Counseling Improve Mortgage Decision-
Making?: Evidence from a Natural Experiment” (2009) (Fisher Coll. of Bus.,Working Paper No.
2008-03-019) 27, available at <http://ssrn.com/abstract=1285603>. In a related study, the
same authors found that financial education reduces mortgage delinquency rates, and attrib-
uted the improved performance to, among other things, the type of mortgage contract
extended to the graduates. See Sumit Agarwal et al., “Learning to Cope:Voluntary Financial
Education Programs and Loan Performance During a Housing Crisis” (2009) Charles A. Dice
Center Working Paper No. 2009-23, available at SSRN: <http://ssrn.com/abstract=1529060>.
See also Annamaria Lusardi, “Household Saving Behavior: The Role of Financial Literacy,
Information, and Financial Education Programs” (2008) (Nat’l Bureau of Econ. Research,
Working Paper No. 13824) 2, available at <http://www.nber.org/papers/w13824> (arguing that
households enter into risky financial contracts due to lack of financial education). See gener-
ally Howard Lax et al., “Subprime Lending: An Investigation of Economic Efficiency”, Hous-
ing Pol’y Debate, 15 (2004), 533, 544–6 (noting that subprime borrowers tend to be less
educated and less sophisticated about the mortgage market).
94. See James M. Lacko and Janis K. Pappalardo, Fed. Trade Comm’n, “Improving Consumer
Mortgage Disclosures: An Empirical Assessment of Current and Prototype Disclosure Forms”
(2007) at ES-6 (demonstrating the limits of mortgage disclosures and noting that many bor-
rowers “did not understand important costs and terms of their own recently obtained mort-
gages. Many had loans that were significantly more costly than they believed, or contained
significant restrictions, such as prepayment penalties, of which they were unaware.”); James M.
Lacko and Janis K. Pappalardo, Fed. Trade Comm’n, “The Effect of Mortgage Broker Com-
pensation Disclosures on Consumers and Competition: A Controlled Experiment” (2004) at
ES-7 (identifying framing effects).
95. Annamaria Lusardi, “Americans, Financial Capability” (2011) NBER Working Paper 17103. See
also Brian Bucks and Karen Pence, “Do Borrowers Know their Mortgage Terms?” (2008) 64
J. Urban Econ., 64 (2008), 218 (finding that comprehension of mortgage terms is low among
borrowers who are exposed to potentially large changes in their mortgage payments).
Other studies have documented specific mistakes that borrowers consist-

ently make. A recent study by Sumit Agarwal, John Driscoll, Xavier Gabaix,
and David Laibson identified persistent mistakes in loan applications that
increased borrowers’ APRs by an average of 125 basis points. Another study,
by Susan Woodward, identified systemic mistakes leading to excessive bro-
ker fees of up to $1,500. In addition, numerous studies have documented
borrowers’ failure to make optimal refinancing decisions. For example, many
consumers fail to exercise refinancing options, thereby ending up with rates
that are significantly higher than the market rate. Other consumers refi-
nance too early, failing to account for the possibility that interest rates will
continue to decline. According to one estimate, these refinancing mistakes
can cost borrowers tens of thousands of dollars or up to 25 percent of the
loan’s value.96
Evidence of rapid defaults—those that occur within six to twelve months
of origination—provides additional support to the behavioral-economics
theory. One explanation for borrowers’ inability to afford the monthly pay-
ments almost from the moment of origination is that they did not fully
understand the extent of the obligations that they were undertaking. Evi-
dence that loan prices are affected by factors unrelated to the risk of non-
payment provides indirect evidence of borrower mistakes. Both data and
testimony by loan officers suggest that many borrowers who might qualify
for prime loans ended up with higher-priced subprime mortgages—an
indication of systematic mistakes. Evidence that borrowers who consider
two or more price dimensions when shopping for a loan end up paying
more for the loan than borrowers who consider only a single price dimen-
sion provides further support for the behavioral explanation.97
96. See Sumit Agarwal et al., “The Age of Reason: Financial Decisions over the Lifecycle” (Octo-
ber 21, 2008) 9–11, available at <http://ssrn.com/abstract=973790> (mistakes in loan appli-
cations); Susan E. Woodward, “Consumer Confusion in the Mortgage Market” (2003),
available at <http://www.sandhillecon.com/pdf/consumer_confusion.pdf> (mistakes leading
to excessive broker fees). On refinancing mistakes—see John Y. Campbell, “Household
Finance”, J. Fin., 61 (2006), 1553, 1579, 1581, 1590; LaCour-Little and Holmes, above note 52,
at 644 (describing the “apparent irrationality on the part of mortgage borrowers, who fail to
default to the extent predicted when house prices fall and fail to prepay to the extent pre-
dicted when interest rates fall”); Agarwal et al., above note 63, at 3 (surveying evidence that
borrowers fail to make optimal refinancing decisions); Agarwal et al., above note 63, at 25, 28
tbl. 5 (many borrowers followed the NPV rule, instead of the optimal-refinancing rule, lead-
ing to substantial expected losses: $26,479 on a $100,000 mortgage, $49,066 on a $250,000
mortgage, $86,955 on a $500,000 mortgage, $163,235 on a $1,000,000 mortgage).
97. On rapid defaults—see Mayer et al., above note 23, at 16 (noting that “2 percent of outstanding
loans in the 2007 vintage were in default within six months of origination, and 8 percent were
in default after 12 months”). On prime borrowers who ended up with subprime loans—see
mortgage s 163
A study by Daniel Bergstresser and John Beshears directly links the lim-
ited sophistication of borrowers to product choice in the mortgage market.
Specifically, Bergstresser and Beshears estimated the relationship between
the ability of respondents to comprehend the financial questions in the
Survey of Consumer Finances (the comprehension rating) and the mort-
gages chosen by these respondents.They found that “during 2004 and 2007,
a change in the comprehension rating from ‘Excellent’ to ‘Poor’ was associ-
ated with a 6.6 percentage point increase in the probability that a home-
owner with a mortgage had an ARM.” Similarly, a study by Morgan Rose
found that non-banks originate disproportionately more loans with prepay-
ment penalties in locales with less financially sophisticated borrowers.98
Industry sources lend further support to the behavioral account. The
National Association of Realtors, in a guide to ARMs and FRMs, writes:
“ARMs are difficult to understand. Lenders have much more flexibility
when determining margins, caps, adjustment indexes, and other things, so
unsophisticated borrowers can easily get confused or trapped by shady
mortgage companies.”99
It seems that few people dispute the fact that at least some borrowers did
not enter into their subprime mortgage contracts with a full understanding
of the costs and benefits associated with these contracts.The FRB, in justify-
ing its new mortgage regulations, referred to borrowers who “unwittingly
Freddie Mac, Automated Underwriting: Making Mortgage Lending Simpler and Fairer for America’s
Families (1996) ch. 5, available at <http://www.freddiemac.com/corporate/reports/moseley/
chap5.htm> (reporting that 10 to 35 percent of subprime borrowers would qualify for lower-
cost conventional loans); Freddie Mac, “Half of Subprime Loans Categorized as ‘A’ Quality,”
Inside B&C Lending, June 10, 1996 (describing a poll of fifty subprime lenders who estimated
that half of subprime borrowers could have qualified for prime loans);“Fannie Mae Has Played
Critical Role in Expansion of Minority Homeownership over Past Decade; Raines Pledges to
Lead Market for African American Mortgage Lending,” Bus. Wire, March 2, 2000, LexisNexis
Academic (noting that up to half of subprime borrowers would qualify for lower-cost conven-
tional loans); Lew Sichelman, “Community Group Claims CitiFinancial Still Predatory,” Origi-
nation News, January 2002, at 25 (reporting that, in 2002, researchers at Citibank concluded that
at least 40 percent of those who were sold high interest rate, subprime mortgages would have
qualified for prime-rate loans); see also Willis, above note 29, at 730; Morgenson, above note 51,
at 9 (recounting that in December 2006, in an agreement with the New York State Attorney
General, Countrywide agreed “to compensate black and Latino borrowers to whom it had
improperly given high-cost loans in 2004”). On paying more for a loan when considering
more price dimensions—see Woodward, above note 96, at 2.
98. See Daniel Bergstresser and John Beshears, “Who Selected Adjustable-Rate Mortgages?: Evi-
dence from the 1989–2007 Surveys of Consumer Finances” (2010) HBS Working Paper 10-083,
4; Morgan J. Rose, “Origination Channel, Prepayment Penalties, and Default”, forthcoming
in Real Estate Economics, available at <http://ssrn.com/abstract=1908375> (using education
level, income, and age to proxy for financial sophistication).
99. See Bergstresser and Beshears, above note 98, at 3.
accept[ed] loans” with terms that they did not fully understand. Likewise, the
CBO noted that “[t]he rise in defaults of subprime mortgages may also
reflect the fact that some borrowers lacked a complete understanding of the
complex terms of their mortgages and assumed mortgages that they would
have trouble repaying.” And HUD’s Report to Congress on the Root Causes
of the Foreclosure Crisis concludes: “Existing evidence suggests that some
borrowers did not understand the true costs and risks of [riskier] loans.”100
D. Market Correction
Individuals are imperfectly informed and imperfectly rational. Yet most
markets work reasonably well despite these imperfections. Several market-
correction mechanisms operate to minimize the effects of imperfect infor-
mation and imperfect rationality. These correction forces were present in
the subprime and Alt-A mortgage markets. As we’ll see in the section that
follows, however, these corrective forces were weak in these markets. For
this reason, borrower mistakes persisted for a prolonged period of time.
Changes in lending practices began only after the subprime market col-
lapsed and legal reforms were implemented.
1. On the Demand Side: Learning by Borrowers

Individuals make mistakes. Most individuals also learn from their mistakes
and learn not to repeat these mistakes.While learning is not absent from the
mortgage market, it is slower. This is because the number of mortgage con-
tracts that individuals sign during the course of a lifetime is small. Inter-
personal learning (learning from others’ mistakes) can compensate for
limited intrapersonal learning (learning from one’s own mistakes), as bor-
rowers share mortgage-related experiences. Interpersonal learning, however,
is not always common enough and detailed enough to eliminate mistakes.
More generally, the evidence shows that learning about financial decisions
is, at best, incomplete.101
100. Truth in Lending, 73 Fed. Reg. 44,522, 44,525–26 (July 30, 2008) (codified at 12 C.F.R.
pt. 226) (FRB); Hearing, above note 1, at 13 (CBO); HUD Report, above note 10, at viii, 22,
52 (HUD).
101. On limited learning when decisions are infrequent—see Truth in Lending, 73 Fed. Reg.
1672, 1676 (January 9, 2008) (codified at 12 C.F.R. pt. 226) (“Disclosures, themselves, likely
cannot provide this minimum understanding for transactions that are complex and that
consumers engage in infrequently.”); Shlomo Benartzi and Richard H. Thaler, “Heuristics
and Biases in Retirement Savings Behavior” J. Econ. Persp., Summer (2007) 81. On limited
learning, and persistent biases, in relatively abstract domains like math and finance—
mortgage s 165
In many markets, effective learning occurs when individuals, aware of

their limitations, seek expert advice.This mechanism also works imperfectly
in the mortgage market. Borrowers commonly seek the advice of mortgage
brokers. These brokers face an incentive structure that prevents them from
being loyal agents of the borrower. (Section 1403 of the Dodd–Frank Act
addresses this concern.) Moreover, the complexity of the subprime mort-
gage contract is such that even so-called experts often get it wrong. For
example, a recent study by Sumit Agarwal, John C. Driscoll, and David
Laibson has shown that available expert advice on refinancing ignores the
option value of postponing the prepayment decision—an omission that can
cost borrowers up to 25 percent of the loan value.102
2. On the Supply Side: Mistake Correction by Sellers

and Reputation Effects
Competing sellers will often have an incentive to correct consumer mistakes.
This can be done, for example, through advertising. While the incentives to
correct consumer mistakes are not always strong in competitive markets, they
are even weaker in imperfectly competitive markets. As explained earlier,
ineffective shopping by borrowers inhibited competition in the subprime
mortgage market. In many markets, a seller’s reputation provides a powerful
deterrent to the abuse of consumers. But, again, reputational forces were
weaker in the subprime mortgage market for several reasons. First, there is
little repeat business, as a single borrower takes few mortgage loans and a
relatively long time passes between loans. Second, many lender organizations
were relatively short-lived. A downside of the securitization innovation was
the opening of the market to fly-by-night originators with little reputation
to lose and insufficient incentives to build a reputation.103
see Thomas Gilovich et al. (eds.), Heuristics and Biases: The Psychology of Intuitive Judgment
(Cambridge University Press, 2002); Keith E. Stanovich, “The Fundamental Computational
Biases of Human Cognition: Heuristics that (Sometimes) Impair Decision Making and Prob-
lem Solving,” in Janet E. Davidson and Robert J. Steinberg (eds.), The Psychology of Problem
Solving (Cambridge University Press, 2003) 291. And real-world evidence of persistent mis-
takes in the mortgage market confirms that learning was limited. See above Sec. IV.C.
102. See Agarwal et al., above note 63, at 24–5.
103. See Engel and McCoy, above note 27, at 2041.The proliferation of small, short-lived sellers is
evidenced by the number of loan originators that have gone out of business during the
recent crisis. See Worth Civils and Mark Gongloff, “Subprime Shakeout: Lenders that Have
Closed Shop, Been Acquired or Stopped Loans,” Wall St. J. Online, <http://online.wsj.com/
public/resources/documents/info-subprimeloans0706-sort.html> (listing eighty loan origi-
nators that closed or filed for bankruptcy between November 2006 and September 2007).
V. Welfare Implications
What are the costs of the identified contractual designs, especially when
understood as a response to borrowers’ imperfect rationality? In this Part,
we’ll look at four potentially substantial costs:
• First, complex, multidimensional contracts hinder competition in the
subprime mortgage market.
• Second, complex and deferred-cost contracts distort the remaining,
weakened forces of competition, leading to excessively high prices on
more salient price dimensions and excessively low prices on less salient
price dimensions.
• Third, these contractual design features increase the likelihood of default
and foreclosure, with all the ensuing costs—to borrowers, lenders, com-
munities, and the economy at large.
• Fourth, the identified contractual designs raise distributional con-
cerns, as they impose disproportionate burdens on weaker—often
minority—borrowers.
Perhaps the costliest result of excessively complex contracts is the inhibited
competition that they foster. As described above, complexity prevents the
effective comparison-shopping that is necessary for vigorous competition.
The market power gained by lenders clearly helps lenders at the expense of
borrowers. But the limited competition also imposes a welfare cost in the
form of inefficient allocation: Borrowers are not matched with the most
efficient lender.
The limits of competition in the subprime mortgage market were
reflected in above-cost pricing. Borrowers were paying origination fees
exceeding the actual costs that these fees allegedly cover by hundreds or
even thousands of dollars. Borrowers were also paying interest rates higher
than what the borrower’s risk profile justified. The most extreme case was
that of borrowers who would have qualified for lower-cost conventional
loans but were nonetheless obtaining high-cost subprime mortgages. The
higher profit margin in the subprime market induced lenders to steer
mortgage s 167
borrowers into subprime loans.104 This problem was explicitly recognized

by the FRB: “[A]n atmosphere of relaxed standards may increase the inci-
dence of abusive lending practices by attracting less scrupulous originators
into the market, while at the same time bringing more vulnerable borrow-
ers into the market. These abuses can lead consumers to pay more for their
loans than their risk profiles warrant.”105
Limited competition allows lenders to set above-cost prices and reap supra-
competitive profits. But even if borrowers engaged in vigorous shopping,
eliminating all supra-competitive profits, there would still be a welfare cost,
because the borrowers’ shopping, while vigorous, would be misguided. Con-
sider again the stylized example of a mortgage contract with a two-dimen-
sional price, a short-term introductory rate, PST, and a long-term rate, PLT .The
two prices affect the two decisions a borrower must make—whether to get
out of the loan at the end of the introductory period, and whether to take the
loan in the first place. An optimal contract will set the two prices to induce
efficient decisions. If borrowers are rational, competition will produce the
optimal contract. This is not the case if borrowers are imperfectly rational. If
borrowers underestimate the costs associated with the long-term rate, PLT ,
competition will focus on the short-term rate, PST , resulting in an inefficient
contract with an excessively low PST and an excessively high PLT .106
There are two adverse welfare implications. First, the excessively high PLT
will lead some borrowers to exit, inefficiently, at the end of the introductory
period. Second, and more importantly, the initial decision to take a loan will
104. On excess fees—see Susan E. Woodward, U.S. Dep’t of Hous. and Urban Dev., A Study of
Closing Costs for FHA Mortgages (2008).The Woodward study found that complexity and mul-
tidimensionality of origination fees prevent effective shopping, hinder competition, and lead
to inflated prices. Id. According to HUD, borrowers are paying excess fees averaging around
$700 per mortgage, and these excess fees can be eliminated by improved disclosure that would
enhance competition. See News Release, U.S. Dep’t of Hous. and Urban Dev., above note 61.
On excess interest—see Engel and McCoy, above note 27, at 2058; Howard Lax et al., “Sub-
prime Lending: An Investigation of Economic Efficiency”, Housing Pol’y Debate, 15 (2004), 533
(arguing that subprime interest rates cannot be justified by risk alone); Stein, above note 51
(valuing the cost to borrowers of excess interest at $2.9 billion). On prime borrowers who
ended up with subprime loans—see above note 97; Morgenson, above note 51 (describing the
steering of prime borrowers into subprime loans at Countrywide).
105. Truth in Lending, 73 Fed. Reg. 1672, 1675 ( January 9, 2008) (codified at 12 C.F.R. pt. 226).
106. See above Part IV.A.
be distorted. While the actual total payments, PST + PLT , will go up to cover
the increased cost generated by the inefficient contractual design, the total
payments as perceived by the borrower will go down.The result is excessive
borrowing.107
This analysis applies to all the examples of cost deferral discussed earlier;
small down payments, high LTVs, escalating payments, and prepayment
penalties. The analysis also applies to the complexity examples, where less-
salient or indirectly specified price dimensions are ignored or underesti-
mated. (PLT corresponds to the less salient, underestimated price dimensions,
and PST corresponds to the more salient price dimensions.) In all of these
cases, imperfect rationality results in price distortions. These distortions
increase total costs and total payments and skew both long-term and short-
term decisions. Most importantly, these distortions increase the actual cost
while reducing the perceived cost of the loan, which leads to an artificially
inflated demand for mortgage financing.
Given the link between the demand for mortgage financing and the
demand for real estate, the identified pricing distortions also contributed to
the housing bubble. HUD, in its Report to Congress on the Root Cause of
the Foreclosure Crisis, observed that nontraditional mortgages (where cost
deferral was most pronounced) enabled buyers to bid on expensive houses,
thus contributing to the housing bubble. The behavioral-market failure in
the mortgage market can therefore be considered an indirect cause of the
financial crisis.108
C. Delinquency and Foreclosure

There is evidence that the identified contractual design features increase
delinquency and foreclosure rates. Deferred-cost contracts are associated
with higher rates of delinquency and foreclosure.These increased delinquency
107. Excessive borrowing would result even absent a contractual-design response—that is, even
under the optimal contract. The contractual design response exacerbates the welfare cost.
108. See HUD Report, above note 10, at 24 (referencing studies by Mian and Sufi (2008), Pav-
lov and Wachter (2008), and Shiller (2007)). See also Andrey Pavlov and Susan Wachter,
“Subprime Lending and Real Estate Prices”, Real Estate Econ., 39 (2011), 1. Cf. Atif R.
Mian and Amir Sufi, “Household Leverage and the Recession of 2007–09”, IMF Econ.
Rev., 58 (2010), 74–117 (showing that “household leverage growth and dependence on
credit card borrowing as of 2006 explain a large fraction of the overall consumer default,
house price, unemployment, residential investment, and durable consumption patterns
during the recession”).
mortgage s 169
and foreclosure rates have been linked to high LTVs, escalating payments,
and prepayment penalties. The FRB, in advocating its new mortgage regu-
lations, acknowledged that “several riskier loan attributes,” including “high
loan-to-value ratio[s]” and “payment shock on adjustable-rate mortgages,”
“increased the risk of serious delinquency and foreclosure for subprime
loans originated in 2005 through early 2007.” HUD in its Report to Con-
gress on the Root Cause of the Foreclosure Crisis concluded: “The sharp
rise in mortgage delinquencies and foreclosures is fundamentally the result
of rapid growth in loans with a high risk of default—due both to the terms
of these loans and to loosening underwriting controls and standards.”109
The welfare costs associated with foreclosure are substantial. FRB Chair-
man Ben Bernanke estimated that, on average, total losses from foreclosure
“exceeded 50 percent of the principal balance, with legal, sales, and mainte-
nance expenses alone amounting to more than 10 percent of principal.” An
109. On the effect of contract design on delinquency and foreclosure—see generally Edward M.
Gramlich, “Subprime Mortgages: America’s Latest Boom and Bust” (Urban Institute Press, 2007)
66–7 (arguing that mortgage contract design is linked to borrower distress); Gene Amromin,
Jennifer Chunyan Huang, Clemens Sialm, and Edward Zhong, “Complex Mortgages” (2010)
FRB of Chicago Working Paper No. 2010-17 (finding that “[b]orrowers with complex mort-
gages experience substantially higher ex post default rates than borrowers with traditional
mortgages with similar characteristics.”). For assessments by the FRB and HUD on the rela-
tionship between contract design and adverse outcomes in the mortgage market—see Truth
in Lending, 73 Fed. Reg. 1672, 1674 (January 9, 2008) (codified at 12 C.F.R. pt. 226) (FRB
assessment); HUD Report, above note 10, at 29 (HUD assessment). On the effect of high
LTVs—see Gerardi et al., above note 21, at 4; Sewin Chan, Michael Gedal, Vicki Been, and
Andrew Haughwout, “The Role of Neighborhood Characteristics in Mortgage Default
Risk: Evidence from New York City” (2011) NYU Working Paper, 12 and tbls. 3, 5. On the
effect of escalating payments, there is evidence that ARMs and Hybrids, which featured
escalating payments, experienced substantially higher rates of delinquency and foreclosure, as
compared to FRMs. See Bernanke, March 2008 Speech, above note 38; Mayer et al., above
note 23, at 8. See also Chan et al, id., 12, 26 and tbl. 3. The effect of interest-rate resets was
more limited than it could have been thanks to the low LIBOR rate at the time, which kept
the reset magnitudes in check. On the effect of prepayment penalties—see Demyanyk and
Van Hemert, above note 1, at 1862 (tbl. 3) (finding positive correlation coefficients on Pre-
payment Penalty in regressions that try to explain default and foreclosure rates); Roberto G.
Quercia, Michael A. Stegman, and Walter R. Davis, “The Impact of Predatory Loan Terms on
Subprime Foreclosures: The Special Case of Prepayment Penalties and Balloon Payments”,
Housing Pol’y Debate, 18 (2007), 311, 337 (finding, based on LP data, that “lengthy”—that is, 3
years or more—prepayment penalties increase foreclosure risk by about 20 percent). The
higher default rates of mortgages with escalating payments (ARMs as compared to FRMs)
and with prepayment penalties were partly due to inherent risk associated with these
deferred-cost features and partly due to poor underwriting standards, which were also associ-
ated with the loan’s deferred-cost features (escalating-payment mortgages enabled lenders to
qualify borrowers based on the low, initial rate; similarly, according to some accounts, prepayment
penalties enabled the lower initial interest rates that qualify riskier borrowers).
industry study that assumes foreclosure losses equal to 37.5 percent of a loan’s
value estimates total subprime foreclosure losses on loans originated between
2004 and 2006 at nearly $29 billion. Substituting Bernanke’s 50 percent fig-
ure for the 37.5 percent assumption, the estimate of foreclosure losses increases
to $38.7 billion. Of this $38.7 billion, the 10 percent (or $7.7 billion) in trans-
action costs—the “legal, sales, and maintenance expenses” that Bernanke
referred to—are clearly welfare costs. The remainder is partly a welfare cost
and partly a welfare-neutral transfer.The transfer component is the “foreclos-
ure discount,” the difference between the market price and the price received
for a foreclosed property. This price discount, while a loss to the lender and
borrower, is a benefit to the buyer of the foreclosed property. The welfare-
cost component is the social loss incurred when a property is left vacant—
until the foreclosure sale and often even after the foreclosure sale. In a
declining real estate market, these vacancy periods are quite long.110
Another category of welfare costs, not included in the preceding esti-
mates, is composed of the negative externalities that foreclosures impose on
neighborhoods and cities. The FRB noted that “[w]hen foreclosures are
clustered, they can injure entire communities by reducing property values
in surrounding areas.”111 Finally, to the extent that foreclosures contributed
110. For the Bernanke estimate—see Bernanke, March 2008 Speech, above note 38. For the industry
study estimates—see Cagan, above note 20, at 69–71. See also Paul S. Calem and Michael LaCour-
Little, “Risk-Based Capital Requirements for Mortgage Loans” (2001) 12 (Bd. of Governors of the
Fed. Reserve Sys., Fin. and Econ. Discussion Series Paper No. 2001–60) (assuming it costs 10 percent
of the unpaid balance to dispose of the foreclosed property and that foreclosure transaction costs
amount to 5 percent of unpaid balance). On the foreclosure discount—see Cagan, above note 20,
at 70 (arguing that foreclosed properties sell at a discount of up to 30 percent).
111. Truth in Lending, 73 Fed. Reg. 44,522, 44,524 (July 30, 2008) (codified at 12 C.F.R. pt. 226).
See also Vicki Been, Dir., Furman Ctr. for Real Estate and Urban Policy, Testimony before
Committee on Oversight and Government Reform Subcommittee on Domestic Policy,
“External Effects of Concentrated Mortgage Foreclosures: Evidence from New York City”
(May 21, 2008) 4–5 (reporting that, in New York, properties adjacent to recent foreclosure
filings sell at a 1.8 percent to 3.7 percent discount); see also William C. Apgar and Mark Duda,
Homeownership Pres. Found., “Collateral Damage: The Municipal Impact of Today’s Mort-
gage Foreclosure Boom” (2005), available at <http://www.995hope.org/content/pdf/
Apgar_Duda_Study_Short_Version.pdf>; Ctr. for Responsible Lending, “Subprime Spillo-
ver: Foreclosures Cost Neighbors $202 Billion: 40.6 Million Homes Lose $5,000 on Average”
(2008), available at <http://www.responsiblelending.org/issues/mortgage/research/
subprime-spillover.html>; U.S. Dep’t of Hous. and Urban Dev. and U.S. Dep’t of the Treasury,
“Curbing Predatory Home Mortgage Lending” (2000) 25 (detailing externalities such as
declines in neighboring property values and increased crime rates); Family Hous. Fund,
“Cost Effectiveness of Mortgage Foreclosure Prevention: Summary of Findings” (1998) 5
(noting foreclosure costs of around $7,000 for borrowers, $2,000 for lenders, and additional
costs of $15,000 to $60,000 on third parties); Dan Immergluck and Geoff Smith,“The Impact
of Single-Family Mortgage Foreclosures on Neighborhood Crime”, Housing Stud., 21 (2006),
851; Engel and McCoy, above note 27, at 2042 n. 12.
mortgage s 171
to the real-estate slump and to the credit crunch, staggering macroeco-

nomic costs should also be considered.
For borrowers, delinquency and foreclosure entail substantial hard-
ship. Borrowers will face higher rates for other credit transactions and
reduced access to credit. They will also lose some or all of their accumu-
lated home equity if the lender forecloses. In addition, the borrower will
have to bear the transaction costs of relocating to another house or
apartment.112
Delinquency and foreclosure also impose costs on lenders. If the net
proceeds from the foreclosure sale are smaller than the outstanding loan bal-
ance, the lender will suffer a loss. Lenders partially compensated for this risk
by increasing the interest rate.113 During the subprime crisis, however, much
of this risk was not priced. The sheer magnitude of the ex post losses—as
reflected in the hundreds of billions of dollars in subprime-related write-
offs by financial institutions—suggests that the risks were not fully accounted
for ex ante. Moreover, SEC investigations, following the collapse of the
subprime market, revealed that at least some lenders had a very poor under-
standing of the risks that they were undertaking.114
In measuring the social cost of foreclosure, it is important to distinguish
between costs borne by borrowers and lenders on the one hand and costs
borne by third parties—neighbors, neighborhoods, and cities—on the other.
For borrowers and lenders, to the extent that the transacting parties were
rational, the ex post cost of foreclosure represents a sour realization of a
mutually beneficial ex ante gamble. Accordingly, we need to worry only
about the imperfectly rational parties who did not secure a positive ex ante
value. Now consider the costs borne by third parties. These costs—negative
externalities imposed by the loan contract—translate into a social cost, even
when both contracting parties are fully rational.
112. Truth in Lending, 73 Fed. Reg. 44,522, 44,524 (July 30, 2008) (codified at 12 C.F.R. pt. 226)
(“The consequences of default are severe for homeowners, who face the possibility of fore-
closure, the loss of accumulated home equity, higher rates for other credit transactions, and
reduced access to credit.”).
113. See Demyanyk and Van Hemert, above note 1, at 1871–3 (finding that high loan-to-value
borrowers increasingly became high-risk borrowers over the past five years, in terms of ele-
vated delinquency and foreclosure rates, and that lenders were aware of this and adjusted
mortgage rates accordingly over time).
114. On the subprime-related write-offs—see above note 12. On the findings from the SEC
investigations—see above note 13.
D. Distributional Concerns
Contractual design can also have distributional effects.While wealthy borrow-
ers were not generally part of the subprime and Alt-A markets, there was still
substantial heterogeneity in the wealth levels of subprime and Alt-A borrow-
ers. Given the complexity of these contracts, wealthier borrowers who could
afford to seek out expert advice were likely to do better than borrowers who
could not afford such advice. The inverse correlation between borrower
wealth and contractual complexity—wealthier borrowers generally got less
complicated prime loans and poorer borrowers generally got more compli-
cated subprime or Alt-A loans—raises another distributional concern.
Evidence that “subprime mortgages [were] concentrated in locations
with high proportions of black and Hispanic residents, even controlling for
the income and credit scores of these Zip codes”115 also raises distributional
concerns. Disparities in financial sophistication and in the ability to effect-
ively comparison-shop led to substantial price variations, even if only
because minority borrowers had fewer options to compare. A study, by Susan
Woodward, found that African-American borrowers paid an additional $415
in fees and Latino borrowers paid an additional $365 in fees. Other price
terms likewise reflected variations. Specifically, “black homeowners [were]
significantly more likely to have prepayment penalties or balloon payments
attached to their mortgages than non-black homeowners, even after con-
trolling for age, income, gender, and creditworthiness.”116
Gender disparities have also been identified: There is some evidence that
women, as a group, received inferior mortgage products.117 Socio-economic
115.. Mayer and Pence, above note 83, at 2.

116.. On the concentration of subprime loans in minority neighborhoods—see Mayer and Pence,
above note 83, at 2. On the limited shopping by minority borrowers—see Michael S. Barr,
Sendhil Mullainathan, and Eldar Shafir, “Behaviorally Informed Home Mortgage Credit
Regulation,” in Eric S. Belsky and Nicolas P. Retsinas (eds.), Understanding Consumer Credit
(Brooking Press, 2009) (“[L]ow-income and minority buyers are the least likely to shop for
alternate financing arrangements....”); Jinkook Lee and Jeanne M. Hogarth, “Consumer
Information Search for Home Mortgages: Who, What, How Much, and What Else?”, Fin.
Services Rev. 9 (2000), 277, 283; Zywicki and Adamson, above note 55, at 55–6. On the higher
fees paid by minority borrowers—see Woodward, above note 104, at ix. On race-based vari-
ations in other price terms—see Michael S. Barr, Jane K. Dokko, and Benjamin J. Keys,“Who
Gets Lost in the Subprime Mortgage Fallout?: Homeowners in Low- and Moderate-Income
Neighborhoods” (April 2008) 2–3, available at <http://ssrn.com/abstract=1121215>; Ruben
Hernandez-Murillo, Andra C. Ghent, and Michael Owyang, “Race, Redlining, and Sub-
prime Loan Pricing” (2011), available at <http://ssrn.com/abstract=1881894>.
117. See John Leland, “Baltimore Finds Subprime Crisis Snags Women,” New York Times, January
15, 2008, at A1; see also Allen J. Fishbein and Patrick Woodall, Consumer Fed. of Am., “Women
mortgage s 173
status also played a role. Borrowers with less income and education were less
likely to know their mortgage terms, implying greater underestimation of
deferred or hidden costs and a diminished ability to effectively shop for bet-
ter terms. Indeed, there is evidence that better-educated borrowers received
better terms on their loans.118
The evidence of bias, however, is not conclusive. In a sample of more
than 75,000 adjustable-rate mortgages, Andrew Haughwout, Christopher
Mayer, and Joseph Tracy found no evidence of adverse pricing by race, eth-
nicity, or gender in either the initial rate or the reset margin. But as the
authors acknowledge, their analysis focuses on interest rates, leaving open
the possibility that bias affects points and fees at loan origination. In addition,
Are Prime Targets for Subprime Lending: Women Are Disproportionately Represented in
High-Cost Mortgage Market” (2006) 1, available at <http://www.consumerfed.org/pdfs/
WomenPrimeTargetsStudy120606.pdf> (finding that women are more likely to receive sub-
prime mortgages than men and that disparity between men and women increases as income
rises); Nat’l Cmty. Reinvestment Coal., “Homeownership and Wealth Building Impeded:
Continuing Lending Disparities for Minorities and Emerging Obstacles for Middle-Income
and Female Borrowers of All Races” (2006) 12–14, available at <http://www.ncrc.org/index
.php?option=com_contentandtask=viewandid=344andItemid=76> (finding that women
received 37 percent of high-cost home loans in 2005, compared with just 28 percent of prime
loans); Prudential Ins. Co. of Am., Financial Experience and Behaviors among Women (2006) 7,
available at <http://www.prudential.com/media/managed/2006WomenBrochure_FINAL.pdf>
(finding that “a majority of financial and investment products are unfamiliar to almost half of
all women”); Annamaria Lusardi and Olivia S. Mitchell, “Planning and Financial Literacy:
How Do Women Fare?” (2008) (Nat’l Bureau of Econ. Research Working Paper No. 13750), avail-
able at <http://ssrn.com/abstract=1087003> (finding that older women display much lower
levels of financial literacy than the older population as a whole); Women in the Subprime
Market, Consumers Union, October 2002, <http://www.consumersunion.org/finance/
women-rpt1002.htm> (attributing some of the disparity both to the instability in women’s
credit status that results from divorce or family medical emergency and to the fact that
women have less wealth than men).
118. On the effects of income and education on knowledge of mortgage terms—see Bucks and
Pence, above note 95, at 3, 20–1, 26. On the effects of education on loan terms—see Wood-
ward, above note 104 (finding that offers made by brokers to borrowers without a college
education are $1,100 higher on average); Thomas P. Boehm and Alan Schlottmann, “Mort-
gage Pricing Differentials across Hispanic, African-American, and White Households: Evi-
dence from the American Housing Survey”, Cityscape: J. Pol’y Dev. and Res., No. 2, (2007), 9,
93, and 105 (finding a negative correlation between education and interest rates); J. Michael
Collins, “Education Levels and Mortgage Application Outcomes: Evidence of Financial Lit-
eracy” (2009) Institute for Research on Poverty Discussion Paper No. 1369–09, available at SSRN:
<http://ssrn.com/abstract=1507276> (finding, based on 2005 Home Mortgage Disclosure
Act data aggregated by Census tract, that tracts with higher rates of college completion
pay lower mean interest rates as reported by lenders for high cost loans); Annamaria
Lusardi, “Financial Literacy: An Essential Tool for Informed Consumer Choice?” (2008) 10
(Nat’l Bureau of Econ. Research, Working Paper No. 14084), available at <http://ssrn.com/
abstract=1149331> (citing a 2003 study by Danna Moore showing that low-literacy borrow-
ers are more likely to purchase high-cost mortgages). Individuals with little education,
women, African-Americans, and Hispanics display particularly low levels of literacy. Id. at 1.
the analysis does not rule out the possibility that borrowers were selectively
steered into subprime mortgage products.119
VI. Policy Implications

As we have seen, borrowers’ imperfect rationality explains several contrac-
tual design features in the subprime mortgage market.The imperfect ration-
ality of borrowers, especially when coupled with contracts designed in
response to such imperfect rationality, produced substantial welfare costs.
Since market forces have proven to be too slow to respond to these prob-
lems, legal intervention should be considered to prevent a recurrence of
these problems. Disclosure regulation is the right place to start. Optimally
designed disclosure, while not a perfect fix, can make a significant differ-
ence. It can help less sophisticated borrowers without significantly restrict-
ing the choices available to more sophisticated borrowers.
A. Disclosing the Total Cost of Credit: The Great Promise

of the APR Disclosure
Perhaps the most important reason to focus on disclosure regulation is
because a disclosure mandate that already exists seems to provide, at least in
theory, an effective response to the behavioral-market failure in the sub-
prime and Alt-A mortgage markets; the APR disclosure, which lenders must
provide under the TILA.The APR is a normalized measure of the total cost
of credit. A lender is required to add up all the different prices and fees that
the borrower is required to pay under the loan contract into a single aggre-
gate amount, the “finance charge,” and disclose this dollar amount. Then, to
facilitate comparison-shopping, the lender is required to translate the finance
charge, from a dollar amount into an APR, and disclose this figure as
well.120
119. Andrew Haughwout, Christopher Mayer, and Joseph Tracy, “Subprime Mortgage Pricing:
The Impact of Race, Ethnicity, and Gender on the Cost of Borrowing” (2009) FRB of New
York Staff Report No. 368.
120. See Truth in Lending Act, Pub. L. No. 90–321, § 107, 82 Stat. 146, 149 (1968) (codified as amended
at 15 U.S.C. § 1606 (2006)) (defining the APR); Truth in Lending Act §§ 121–31, 82 Stat. at
152–57 (codified as amended at 15 U.S.C. §§ 1631–49 (2006)) (requiring disclosure of the APR).
See also Renuart and Thompson, above note 19, at 217 (“Congress designed the APR to be the
single number that consumers should focus upon when shopping for credit.”).
mortgage s 175
The importance of total cost disclosures, and of the APR specifically, has
been reaffirmed by the Mortgage Reform and Anti-Predatory Lending Act,
enacted as Title XIV of the Dodd–Frank Act of 2010. Section 1419 of the
Dodd–Frank Act requires several quasi-total-cost disclosures: Creditors must
disclose the total amount of interest the consumer will pay over the life of
the loan, the aggregate amount of fees paid to the mortgage originator in
connection with the loan, and the amount paid for settlement services.
These are called “quasi-total-cost disclosures” because they aggregate costs
in categories (interest, fees, and settlement costs) instead of disclosing a sin-
gle, real total-cost figure.
The new disclosure forms being developed by the Consumer Financial
Protection Bureau (CFPB) to comply with a direct Congressional mandate in
the Dodd–Frank Act secure an important place for the APR, which is theo-
retically a real total-cost disclosure.The forms being tested by the CFPB retain
the APR as a key total-cost disclosure, and include a “Comparisons” section,
featuring the APR, together with Estimated Closing Costs. They even feature
this notice: “Use this information to compare this loan with others.”121
Real total-cost disclosures, and specifically the APR, should serve as a power-
ful antidote to the effects of imperfect rationality for two reasons. First, the APR
would seem to offer an effective response to the complexity and multidimen-
sionality of the subprime mortgage contract. Lenders are required to calculate
and disclose the total loan cost to the borrower. With this standard metric at
hand, borrowers should be able to compare the total cost of two different, com-
plex loan contracts. Collecting all the rates and fees and folding them into a
single aggregate price, the APR renders the borrowers’ cognitive deficiencies—
limited attention, memory, and processing ability—irrelevant. Second, the APR
should provide an effective remedy to the myopia and optimism that give rise
to deferred-cost contracts. Since the APR is a composite of short-term and
long-term interest rates, capturing both long-term costs and short-term bene-
fits, it should reveal the false allure of deferred-cost contracts.122
By overcoming, or bypassing, the imperfect rationality of borrowers, the
APR disclosure should also discourage many of the contractual design
121. See Dodd–Frank Act, Sec. 1032(f) (directing the CFPB to develop a new disclosure form); CFPB,
Know Before You Owe Initiative <http://www.consumerfinance.gov/knowbeforeyouowe/>.
122. On the APR as an antidote to complexity—see Renuart and Thompson, above note 19, at
214 (arguing that a comprehensive, fee-inclusive APR will help imperfectly rational consum-
ers who cannot aggregate the multiple fees on their own). On the APR as a composite of
short-term and long-term interest rates—see 12 C.F.R. § 226.17 (2008); Official Staff Com-
mentary § 226.17(c)(1)–(10) (2008).
features we’ve already explored. Consider complexity and specifically the

proliferation of “junk” fees. Adding non-salient fees was beneficial to the
lender because imperfectly rational borrowers ignored them. But if these fees
are included in the APR and borrowers shop for low APRs, then the incen-
tive to pile up more fees disappears.123 Similarly, cost deferral was an attractive
strategy for lenders because myopic and optimistic borrowers placed insuf-
ficient weight on the long-term costs. If borrowers look to the APR for
guidance, and the APR calculation affords appropriate weight to both short-
term and long-term costs, lenders will have no incentive to defer costs.
There is already evidence that the APR disclosure can work. Many bor-
rowers know to look for the APR and comparison-shop based on that
disclosure. This has led to enhanced competition and reduced rates.124 There
is even evidence that the APR succeeded in fighting imperfect rationality.
Specifically, Victor Stango and Jonathan Zinman show that the most biased
consumers—consumers who substantially underestimate the APR correspond-
ing to a given payment stream—do not overpay for credit when borrowing in
markets where TILA disclosures are consistently made, while these same types
of consumers pay 300 –400 basis points more in interest than less-biased con-
sumers do in markets where TILA disclosures are not made consistently.125
123. See Bd. of Governors of the Fed. Reserve Sys. and U.S. Dep’t of Hous. and Urban Dev., Joint
Report to the Congress Concerning Reform to the Truth and Lending Act and the Real Estate Settle-
ment Procedures Act (1998) 9, available at <http://www.federalreserve.gov/boarddocs/press/
general/1998/19980717/default.htm> (“[T]he APR concept deters hidden or ‘junk’ fees to
the extent that the fees must be included in the APR calculation.”).
124. On consumers’ use of the APR—see Lee and Hogarth, above note 116, at 286 (finding that
78 percent of homeowners who refinanced their homes report comparison shopping on the
basis of the APR); Jinkook Lee and Jeanne M. Hogarth, “The Price of Money: Consumers’
Understanding of APRs and Contract Interest Rates”, J. Pub. Pol’y and Marketing, 19 (1999),
66, 74 (reporting that more than 70 percent of the population reports using the APR to shop
for closed-end credit); Renuart and Thompson, above note 19, at 189 (“TILA disclosures
have been remarkably effective in educating consumers to pay attention to the APR as a key
measure of the cost of credit.”). On the competition-enhancing role of the APR—see Rep.
No. 96-368, at 16 (1979), reprinted in 1980 U.S.C.C.A.N. 236, 252 (crediting TILA with
increasing consumer awareness of annual percentage rates and with a substantial reduction of
the market share of creditors charging the highest rates); Randall S. Kroszner, Governor, Bd.
of Governors of the Fed. Reserve Sys., Speech at the George Washington University School
of Business Financial Services Research Program Policy Forum: “Creating More Effective
Consumer Disclosures” (May 23, 2007), available at <http://www.federalreserve.gov/
newsevents/speech/Kroszner20070523a.htm> (stating that TILA disclosure requirements
and specifically the APR disclosure “are generally believed to have improved competition
and helped individual consumers” (citing Bd. of Governors of the Fed. Reserve Sys., Annual
Percentage Rate Demonstration Project (1987))).
125. Victor Stango and Jonathan Zinman, “Fuzzy Math, Disclosure Regulation and Credit Mar-
ket Outcomes” (2007) Tuck Sch. of Bus. Working Paper No. 2008-42, available at <http://ssrn.
com/abstract=1081635>.
mortgage s 177
B. The Failure of the APR Disclosure

Despite the achievements of the APR disclosure, there is broad consensus that
the APR has not lived up to its great potential and that the current disclosure
regime has failed to protect borrowers and ensure an efficient market—espe-
cially in the subprime and Alt-A markets.126 Why? The answer lies in several
defects that prevented the APR from living up to its great promise.
First, the APR was often disclosed too late. Lenders were not required to
disclose a binding APR—that is, an APR that they cannot change after the
disclosure—until consummation of the loan transaction (closing). In pur-
chase loans, lenders were required to disclose a good-faith estimate of the
APR three days after receiving a loan application. But lenders were not
bound by this estimate.Thus, borrowers could not rely on it when shopping
for loans. In addition, the estimated APR was sometimes provided only after
a substantial application fee was paid. Borrowers who were understandably
reluctant to pay numerous application fees could not use the estimated APR
for comparison shopping. The situation was even worse with refinance
loans, where lenders were not required to provide any disclosure before
closing. Disclosing a binding APR only at closing discourages APR-based
comparison shopping. Few borrowers who reached the closing stage would
then, after finally learning the APR, refuse to sign the loan documents and
start shopping again. (Note that to compare the APR on one loan with the
APR on a competing loan, the borrower would have to reach the closing
stage with the second loan as well.)127
The second reason the APR failed to live up to its potential is that while
purporting to provide a total-cost-of-credit measure, the APR actually
excludes numerous price dimensions, such as title insurance fees, title exam-
ination fees, property survey fees, appraisal fees, credit report fees, document
126. The evidence showing the success of the APR is limited to the prime market. See above
notes 124–25; see also Patricia A. McCoy, “Rethinking Disclosure in a World of Risk-Based
Pricing”, Harv. J. on Legis., 44 (2007), 123, 126, 138–9 (noting robust competition in the prime
market and that TILA disclosures effectively facilitate this competition). On the general fail-
ure of the TILA disclosure regime in the nonprime segments—see, for example, GAO AMP
Report, above note 26, at 21 (noting that current disclosure requirements “are not designed
to address more complex products such as [Alternative Mortgage Products]”); Edward L.
Rubin, “Legislative Methodology: Some Lessons from the Truth-in-Lending Act”, Geo. L.J.,
80 (1991), 233, 236 (noting that shopping for credit is limited to “upscale consumers who
would manage perfectly well without [the] benefit of [the TILA disclosures]”).
127. See McCoy, above note 126, 137–43. See also Willis, above note 29, at 749–50. FTC Com-
ment, above note 31, at 11–12. The exception was HOEPA loans, where binding early
disclosures were required. See McCoy, above note 126, at 141.
preparation fees, notary fees, flood and pest inspection fees, seller’s points,
prepayment penalties, and late fees. By excluding these price dimensions,
the APR underestimates the total cost of the loan. Moreover, this exclusion
invites strategic pricing by lenders. When certain price dimensions are
excluded from the APR, lenders will benefit from shifting costs to these
excluded dimensions.128 These problems undermine the effectiveness of the
APR. Because the APR does not measure the total cost of credit, borrowers
are less likely to focus on the APR. Borrowers who nevertheless use the
APR for comparing loans may well end up with a product that, while
boasting a lower APR, costs more overall.
The third defect is that the APR disclosure fails to account for the pre-
payment option—an option that has dramatically affected the values of sub-
prime and Alt-A loans in the recent mortgage-lending expansion. The
prepayment option can have a significant effect on a loan’s value, even for
traditional, prime loans.129 The effect on subprime and Alt-A loans that were
taken with intent to prepay before the end of the low-rate introductory
period can be much greater. Consider a 2–28 hybrid for $150,000 with a
monthly payment of $1,000 for the first two years and a monthly payment
of $1,500 for the remaining 28 years. The APR on this loan, ignoring the
prepayment option, is 10.74 percent. Assuming that before the 2–28 mort-
gage resets, the borrower can refinance into a 30-year FRM with a $1,000
monthly payment, the effective APR is 7.19 percent.130 The effect of an
attractive prepayment option is significant.
128. For price dimensions excluded from the APR—see Comptroller of the Currency, Truth in
Lending: Comptroller’s Handbook (2006) 98, available at <http://www.occ.treas.gov/handbook/
til.pdf> (showing that the APR does not include late fees, title insurance fees, title examination
fees, property survey fees, appraisal fees, credit report fees, document preparation fees, notary
fees, flood and pest inspection fees, and seller’s points); Willis, above note 29, at 744, 747, 750
(noting APR includes origination fees and points, but not interest rate escalations, prepayment
penalties, late fees, title insurance, and application, appraisal, and document preparation fees)
. On the resulting strategic pricing—see Renuart and Thompson, above note 19, at 185, 221;
Zywicki and Adamson, above note 55, at 71.
129. On the exclusion of the prepayment option—see HUD-Treasury Report, above note 19, at
66 (noting that “the APR does not account for an early payoff ”).This problem persists in the
new disclosure forms that are being considered by the CFPB. The APR disclosure, in these
forms, is even followed by an express statement: “This rate expresses your costs over 30 years.”
See CFPB, Know Before You Owe Initiative, <http://www.consumerfinance.gov/
knowbeforeyouowe/>. On the effect of the prepayment option on loan value—see Agarwal
et al., above note 63, at 28 (calculating a 26.8 percent impact on a $100,000 mortgage for
using the wrong rule to make prepayment decisions; the impact of ignoring the prepayment
option altogether may well be larger).
130. The actual (no prepayment) and effective (with prepayment) APRs were calculated using
APRWIN (Ver. 6.1.0).
mortgage s 179
Moreover, since the prepayment option affects different contractual

designs differently, an APR that ignores the prepayment option can skew
the comparison among different loan products. The prepayment option
might render the APR disclosure misleading even with simple loan con-
tracts. Comparing two loans, Loan A and Loan B, the APR on Loan A can
be lower, reflecting a lower total cost of credit absent prepayment. But with
prepayment, the total cost of Loan B may well be lower.131 This problem is
exacerbated when complex contracts include a set of varying terms that
interact differently with the prepayment option.
The term that most obviously affects the value of the prepayment option
is the prepayment penalty. Many have expressed concerns about prepayment
penalties, and their use has been substantially curtailed by the Dodd–Frank
Act.132 The fear is that since prepayment penalties are not incorporated into
the APR, borrowers will underestimate their effect on the total cost of the
loan.133 These concerns are valid but address only one aspect of the problem.
Those critical of prepayment penalties focus on the penalties that borrowers
actually pay and on borrowers’ underestimation of these payments. They
ignore the effects of prepayment penalties on the value of the prepayment
option. Moreover, prepayment penalties reduce the ex ante value of the
prepayment option even when they are not paid ex post.
An APR that ignores the prepayment option will play a reduced role in
the shopping decisions of perfectly rational borrowers. It will play an even
less significant role in the shopping decisions of imperfectly rational bor-
rowers who overestimate the value of the prepayment option. Unfortu-
nately, this prepayment flaw in the APR calculation enabled even honest
brokers and loan officers to deflect borrowers’ attention from the APR dis-
closure. For example, the APR on a deferred-cost loan could be much
higher than the initial teaser rate. Loan originators wanted borrowers to
131. In particular, by ignoring the prepayment option, the APR underestimates the importance of
origination fees (those that are included in the APR calculation) that accrue at closing. See
Renuart and Thompson, above note 19, at 231.This may provide another explanation for the
proliferation of origination fees.
132. Pub. L. 111–203, Sec. 1414. See also Truth in Lending, 73 Fed. Reg. 44,522, 44,551 (July 30,
2008) (codified at 12 C.F.R. pt. 226).
226) (“Subprime loans are also far more likely to have prepayment penalties. Because the
annual percentage rate (APR) does not reflect the price of the penalty, the consumer must
both calculate the size of the penalty from a formula and assess the likelihood of moving or
refinancing during the penalty period. In these and other ways, subprime products tend to
be complex for consumers.”).
focus on the low teaser rate and not on the high APR. These brokers and
loan officers could truthfully tell borrowers that they are likely to prepay
and exit long before the nominal thirty-year loan period ends and that they
should therefore pay little attention to an APR that assumes thirty years of
loan payments.
The APR disclosure has failed. It was often disclosed too late to help
borrowers choose between different loan products. By excluding numerous
price dimensions and by ignoring the prepayment option, the APR has
failed to live up to its declared purpose of providing an accurate total-cost-
of-credit measure. As a result, borrowers abandoned the APR, and it ceased
to be the focal point of comparison-shopping in the subprime mortgage
market. The resulting cost to borrowers and to society more generally was
substantial.
As mentioned earlier, the APR has the potential to ameliorate the effects
of imperfect rationality, but it can effectively respond to the imperfect
rationality of borrowers only if imperfectly rational borrowers rely on the
APR. Many borrowers, however, did not.
C. Fixing the APR Disclosure

Given the potential of the APR disclosure to compensate for the imperfect
rationality of borrowers, it should be a priority for policymakers to fix the
APR’s problems. In fact, the timing problem has already been addressed—
and partially solved—by recent legal reforms. Specifically, new FRB regula-
tions require lenders to disclose an APR within three days after the loan
application has been submitted and before any fees are charged, for both
purchase and refinance loans. Further, the Housing and Economic Recov-
ery Act requires lenders to disclose an updated APR three days before
consummation of the loan transaction, in case the previously disclosed APR
“is no longer accurate.”134
These recent statutory and regulatory responses reduce but don’t solve
the timing-of-disclosure problem. Two issues remain: First, lenders can still
disclose a low APR after receiving an application, and then disclose a higher
APR later on. Borrowers will be wary of using the application-stage APR
for comparison-shopping, since this APR can change. Three days before
134. See Truth in Lending, 73 Fed. Reg. 44,522, 44,590–92 (July 30, 2008) (codified at 12 C.F.R.
pt. 226); Housing and Economic Recovery Act of 2008, Pub. L. No. 110–289, § 2502(a), 122
Stat. 2654, 2855–57 (codified at 15 U.S.C. § 1638(b)(2)).
mortgage s 181
closing, the time when an updated APR is provided, may already be too late
for effective comparison-shopping. Second, the enforcement of these
improved timing-of-disclosure rules is imperfect. Specifically, several appel-
late courts have interpreted TILA’s civil liability section as precluding statu-
tory damages for timing-of-disclosure violations. The borrower would thus
have to claim actual damages and prove detrimental reliance—a substantial
barrier to recovery.135 While Congress and the FRB should be commended
for reducing the timing-of-disclosure problem, still more can and should be
done. Disclosure of a binding APR should be required at an earlier time,136
and the civil liability provisions of TILA should be strengthened.
The second major APR problem, under-inclusiveness, has not been
addressed. The purpose of the APR was to provide a uniform total-cost-of-
credit measure. The current APR excludes numerous price dimensions and
thus fails to present the total cost of credit.The analysis in this chapter lends
further support to proposals, most recently by Elizabeth Renuart and Diane
Thompson, to create a more inclusive APR.137 Several price dimensions
currently excluded from the APR definition can be easily added; others can
only be added at a cost. For example, adding the price of truly optional
services to the APR would generate several APRs for a single mortgage,
potentially confusing rather than assisting borrowers. Adding contingent
135. On the concern that lenders will set a low APR and then increase it three days before clos-
ing—see Kathleen C. Engel and Patricia A. McCoy, “A Tale of Three Markets: The Law and
Economics of Predatory Lending”, Tex. L. Rev., 80 (2002), 1255, 1269 (noting that lenders
face no liability for errors in the Good Faith Estimate (GFE), including the GFE of the APR).
Moreover, it is not clear from the language of the statute that lenders cannot change the APR
again between the time of the updated disclosure (three days before closing) and consumma-
tion. On the narrow interpretation of TILA’s civil liability section—15 U.S.C. § 1640
(2006)—see, e.g., Dykstra v.Wayland Ford, Inc., 134 F. App’x 911 (6th Cir. 2005); Baker v. Sunny
Chevrolet, Inc., 349 F.3d 862 (6th Cir. 2003); Brown v. Payday Check Advance, Inc., 202 F.3d 987
(7th Cir. 2000); In re Ferrell, 358 B.R. 777 (B.A.P. 9th Cir. 2006). Other courts have adopted a
more expansive interpretation of TILA’s civil liability provisions. See, e.g., Bragg v. Bill Heard
Chevrolet, Inc., 374 F.3d 1060 (11th Cir. 2004).
136. See HUD-Treasury Report, above note 19, at 67 (proposing that originators be required to
provide an accurate, within a prescribed tolerance, Good Faith Estimate of, among other
things, the APR). It should be recognized, however, that locking in an APR at an earlier time
would place greater interest rate risk on the lender and that this added risk would be, at least
partially, passed on to borrowers. Borrowers who need the APR as a focal point for compar-
ison-shopping should be willing to accept these consequences. Cf. McCoy, above note 126,
at 138 (arguing that similar rate lock-ins are common in the prime market even though lend-
ers are not required to disclose a binding APR).
137. Renuart and Thompson, above note 19; see also HUD-Treasury Report, above note 19, at 69
(proposing that the law be amended “to require that the full cost of credit be included in the
APR”); Eskridge, above note 19 (proposing a more inclusive APR more than twenty years ago).
prices, such as late fees and prepayment penalties, imposes a different cost.
These prices can only be incorporated into the APR by estimating the
average probability that the fee-triggering contingency will materialize.
An APR based on this estimated average would be inaccurate for many
borrowers. Of course, the current APR is similarly inaccurate for many bor-
rowers, since it, in effect, assumes a zero probability of triggering these
contingent fees. While a more inclusive APR is warranted, for some price
dimensions the inclusion decision requires a careful cost-benefit analysis.
The third APR problem is the ignored prepayment option. This also has
not been addressed by policymakers and has even escaped the attention of
commentators. When borrowers expect, rationally or irrationally, to avoid
high long-term costs by refinancing their mortgage, they will ignore an
APR that does not include the prepayment option. It is, therefore, useful to
consider incorporating the prepayment option into the APR calculation.To
be sure, accounting for the possibility of prepayment is not an easy exercise.
The likelihood and timing of prepayment would have to be estimated, as
would the expected terms of the refinance loan. These estimates would
need to be based on projections of future house prices (for each Metropoli-
tan Statistical Area) and interest rates.These future market conditions would
then need to be combined with estimated borrower and loan characteris-
tics, such as future FICO score, future income, and future LTV, to estimate
the refinancing options that would be available to the specific borrower.138
These estimates and projections would necessarily be based on a series of
assumptions. While the use of assumptions is not new to disclosure regula-
tion, it should be recognized that some degree of arbitrariness in the choice
of assumptions is inevitable and that the chosen assumptions will not per-
fectly reflect every borrower’s situation. The difficulties of generating accu-
rate projections should not be exaggerated. The mortgage industry already
employs sophisticated valuation algorithms to arrive at projections tailored
to specific home and loan characteristics.139 An APR disclosure that uses
138. Estimating the future LTV is particularly complicated. This estimate would be based on the
current LTV, the contractually specified payment stream, the prepayment penalty—which
would need to be financed by the new loan—and the projected future house value.
139. Projections and forecasts are commonly used in the industry. See, e.g., Cagan, above note 20;
Sherlund, above note 125, at 11 (“I draw house price, interest rate, and unemployment rate
forecasts from Fannie Mae’s and Freddie Mac’s June 2008 monthly economic outlooks. . . . ”);
cf.W. Miles, “Boom-Bust Cycles and the Forecasting Performance of Linear and Non-Linear
Models of House Prices”, J. Real Est. Fin. and Econ., 36 (2008), 249 (comparing the power of
competing models to predict house prices). Futures markets can be used to help predict price
trajectories. And sophisticated valuation algorithms can be used to more closely tailor predic-
mortgage s 183
these projections to account for the prepayment option will thus reduce the
information asymmetry between lenders and borrowers. More importantly,
this disclosure could restore borrower confidence in the APR, thus harness-
ing the potential of the APR to counteract the effects of imperfect
rationality.140
It is worth reminding ourselves that even an optimally designed APR
will not be perfect. It is impossible to fully capture the multidimensionality
of a mortgage loan in a one-dimensional metric. This inevitable limitation,
however, does not detract from the social value of the APR disclosure.
Sophisticated borrowers who can deal with the complexity and multi-
dimensionality will not rely solely on the APR. Those who rely solely, or
mainly, on the APR will be the less sophisticated borrowers who, absent the
APR disclosure, would rely on an even less accurate proxy.141
Conclusion
During the subprime boom years, subprime and Alt-A mortgage contracts
were complex, multidimensional contracts that often deferred costs into the
future. This contractual design can be explained as a market response to the
imperfect rationality of borrowers. The welfare costs of this market failure
were substantial: Competition was both hindered and distorted, resulting in
an inefficient allocation of resources. Default and foreclosure rates increased,
tions to specific homes and specific loans. See Cagan, above note 20, at 5 (describing the
valuation algorithms). See also Philip Bond, David K. Musto, and Bilge Yilmaz, “Predatory
Mortgage Lending,” (2008) FRB of Philadelphia Working Paper No. 08-24, available at SSRN:
<http://ssrn.com/abstract=1288094> (describing the large amounts of information that
lenders have, including information on the performance of specific mortgage types when
taken by borrowers with specific characteristics).
140. The proposed disclosure would also assist rational borrowers. Currently, these borrowers must
calculate the value of the prepayment option (or the probability of facing an attractive prepay-
ment option) on their own.This is a costly exercise. And some borrowers may decide to forgo
the exercise. The proposed disclosure would save the calculation costs or, for those borrowers
who would forgo the exercise, reduce uncertainty about the prepayment option.
141. The limits of the APR, even when optimally designed, warrant consideration of supplemen-
tary approaches. For example, the CFPB could sponsor a web-based mortgage search tool.
This tool would ask the borrower for information relevant to loan underwriting and then
provide a list of best options (from the best lenders), where the best options, or at least some
of them, would not necessarily be picked solely by the APR. Cf. John Lynch, Consumer Infor-
mation Processing and Mortgage Disclosures (2008), available at <http://www.ftc.gov/be/
workshops/mortgage/presentations/Lynch_John.pdf> (proposing a “personalized screening
agent website for best alternatives in region”).
imposing costs on borrowers, lenders, neighborhoods, cities, and the econ-

omy at large. Distributional problems also surfaced.
In this chapter, we explored how the outcome in the subprime and Alt-A
markets can be improved by revitalizing the APR disclosure. The APR, by
providing a common total-cost-of-credit measure, can serve as an effective
antidote to imperfect rationality. The APR can only do so, however, if bor-
rowers focus on the APR when choosing among different mortgage prod-
ucts. In the subprime and Alt-A markets, borrowers largely abandoned the
APR. This can change. Borrowers will again rely on the APR if it is dis-
closed early enough and if it is redesigned to provide a comprehensive total-
cost-of-credit measure.To this end, Congress and the CFPB should minimize
the number of price dimensions that are excluded from the APR definition
and consider incorporating the prepayment option into the APR calcula-
tion. These proposals, if successful in restoring borrower confidence in the
APR, will allow the subprime and Alt-A markets to benefit from the APR’s
unique ability to combat imperfect rationality.
It should be noted that the Dodd–Frank Act, in addition to taking impor-
tant steps towards improving the mortgage disclosure regime, also moves
beyond disclosure. In particular, the Act targets some of the cost-deferral
features identified above: The Act requires lenders to verify a borrower’s
ability-to-repay (Sec. 1411) and sets a safe harbor for the ability-to-repay
requirement—the qualified mortgage, which cannot include certain
deferred-cost features (Sec. 1412). Furthermore, the Act severely restricts the
use of prepayment penalties (Sec. 1414). These reforms reflect Congress’s
recognition that imperfectly rational borrowers might underestimate the
importance of deferred costs.
4
Cell Phones
Introduction
The cellular service market is an economically significant market that has
substantially enhanced consumer welfare. From 1990 to 2009, the U.S. mar-
ket grew from 5 million subscribers to 291 million subscribers. At the time
of this writing, 93 percent of Americans have a cell phone, and an increasing
number of households have given up their landlines and rely entirely on
wireless communications. Annual revenues of the four national carriers—
AT&T, Verizon, Sprint, and T-Mobile—total over $180 billion.
While acknowledging these successes and welfare benefits, the focus of
this chapter is on the failures of this market. We’ll see how carriers design
their contracts in response to the systemic mistakes and misperceptions of
their customers. In doing so, they impose welfare costs on consumers, reduc-
ing the net benefit that consumers derive from wireless service. We’ll focus
on three design features common to most cellular service contracts:
• three-part tariffs;
• lock-in clauses; and
• sheer complexity.
As you have no doubt noticed, a major theme of this book is that the inter-
action between consumer psychology and market forces results in contracts
that feature complexity and deferred costs. Lock-in clauses and three-part
tariffs together generate cost deferral. Lock-in clauses enable bundling of
handsets and cellular service. This bundling allows carriers to offer free or
subsidized phones—an upfront benefit—recouping costs at the back end
through the price of cellular service. This cost deferral is motivated by con-
sumers’ demand for short-term perks and their relative inattentiveness to
long-term costs.The underestimation of long-term costs is amplified by the

three-part tariff, which responds to and exacerbates the effect of mispercep-

tions that lead consumers to underestimate the cost of cellular service. Sheer
complexity is the third of the three design features that contribute to mar-
ket failure.
A. Three Design Features

The basic pricing scheme of the common cellular service contract is a
three-part tariff comprising (1) a monthly charge, (2) a number of voice
minutes that the monthly charge pays for, and (3) a per-minute price for
minutes beyond the plan limit.The three-part tariff is a rational response by
sophisticated carriers to consumers’ misperceptions about their cell phone
usage. Consumers choose calling plans based on a forecast of future use pat-
terns. The problem is that many consumers do not have a very good sense
of these use patterns—some underestimate whereas others overestimate
their future usage. The three-part tariff is advantageous to carriers because
it exacerbates the effects of consumer misperception, leading consumers to
underestimate the cost of cellular service.
The overage-fee component of the three-part tariff targets the under-
estimators.These consumers underestimate the probability of exceeding the
plan limit and incurring an overage fee. As a result, they underestimate the
total cost of the cellular service. The other components of the three-part
tariff, the monthly charge and the fixed number of minutes that come with
it, target the over-estimators.These consumers think that they will use most
or all of their allotted minutes. They therefore expect to pay a per-minute
price equal to the monthly charge divided by the number of allotted min-
utes. In fact, the over-estimators use far fewer minutes and end up paying a
much higher per-minute price. In this way, then, over-estimators also under-
estimate the cost of cellular service.
Carriers seem to be aware of these misperceptions. As a pricing manager
a top U.S. cellular phone carrier explained, “people absolutely think they
know how much they will use and it’s pretty surprising how wrong they
are.”1 The prevalence of consumer misperception can be empirically con-
firmed by using a unique dataset of subscriber-level monthly billing and
usage information for 3,730 subscribers at a single wireless provider. These
1. Michael Grubb, “Selling to Overconfident Consumers”, Amer. Econ. Rev., 99 (2009), 1770, 1771
(note 2).
c e l l phone s 187
data enable calculation of not only the total cost of wireless service under
each consumer’s chosen plan, but also the total amount that the consumer
would have paid had he chosen other available plans. Thus, one can determine
the plan that best fits actual cell phone usage. The data show that over 65
percent of consumers chose the wrong plan. Some chose plans with an
insufficient number of allotted minutes, whereas others chose plans with
an excessive number of allotted minutes. Subscribers exceeded their minute
allowance 17 percent of the time by an average of 33 percent, suggesting
underestimation of use. And, during the 81 percent of the time when the
allowance was not exceeded, subscribers used only 47 percent of their
minute allowance on average, suggesting overestimation.
In addition to the three-part tariff pricing structure, most calling plans
come with a free or substantially discounted phone and a long-term con-
tract with an early termination fee (ETF) that effectively locks the con-
sumer in for a substantial time period—typically two years. These lock-in
clauses and the accompanying ETFs can also be explained as a market
response to the imperfect rationality of consumers. Imperfectly rational
consumers underestimate the cost of lock-in, since they underestimate the
likelihood that switching providers will be beneficial down the road. Switch-
ing providers may be beneficial, for example, if current service is not as
good as promised, monthly charges are higher than expected (due to the
misperception of use levels discussed above), or another carrier is offering a
better deal.
The lock-in that is enforced by the ETF also facilitates the common
practice of bundling phones and service.The long-term revenue stream that
lock-in guarantees enables carriers to offer free or subsidized phones.
Rational consumers, knowing that they will pay for this “free” phone in the
long term, would not be enticed by a free-phone offer. Imperfectly rational
consumers, on the other hand, discount the long-term cost and seek out
“free” phone offers.
Finally, the third design feature that contributes to the behavioral market
failure is the sheer complexity of the cell phone contracts. Cellular service
contracts are complex and multidimensional. Choosing among numerous
contracts can be a daunting task. The three-part tariff itself is complex.
Lock-in clauses and ETFs add further complexity. In addition, the true cost
of a calling plan depends on numerous other features. For example, most
plans offer unlimited night and weekend calling, but carriers offer different
definitions of “night” and “weekend.”Also, consumers must choose between
unlimited in-network calling, unlimited calling to five numbers, unlimited

Walkie-Talkie, rollover minutes, and more. Finally, different carriers offer
different ranges of handsets, handset subsidies vary, and so on. Complexity is
further increased when family plans are added to the mix, data services are
added to voice services, prepaid plans are considered in addition to postpaid
plans, and so forth. According to one industry estimate, the cellular service
market boasts over 10 million plan and add-on combinations.
This level of complexity can itself be viewed as a contractual design fea-
ture that responds to the imperfect rationality of consumers. Complexity
allows providers to hide the true cost of the contract. Imperfectly rational
consumers do not effectively aggregate the costs associated with the differ-
ent options and prices in a cell phone contract. Inevitably, consumers will
focus on a subset of salient features and prices, and ignore or underestimate
the importance of the remaining non-salient features and prices. In response,
providers will increase prices or reduce the quality of the non-salient fea-
tures.This, in turn, will generate or free up resources for intensified compe-
tition on the salient features. Competition forces providers to make the
salient features attractive and the salient prices low. This can be achieved by
adding revenue-generating, non-salient features and prices. The result is an
endogenously derived high level of complexity and multidimensionality.
Interestingly, consumer learning can exacerbate the problem. When con-
sumers learn the importance of a previously non-salient feature, carriers
have a strong incentive to come up with a new one, further increasing the
level of complexity.
B. Rational-Choice Explanations?
Before we can draw normative and prescriptive implications from these
behavioral theories, we must consider whether the more traditional rational-
choice model can explain the same design features. If the rational-choice
model comes up short, then we have good reason to appeal to behavioral
economics to assess the appropriate policy response.
The leading rational-choice explanation for three-part tariffs views these
tariffs as mechanisms for price discrimination or market screening among
rational consumers with different ex ante demand characteristics.The price-
discrimination argument rests on specific assumptions about the distribu-
tion of consumer types—assumptions that are not borne out in the cell
c e l l phone s 189
phone market. With the distribution of types that we actually observe, pro-
viders selling to rational consumers would not offer three-part tariffs.
Lock-in clauses can arise when consumers are rational. This happens
when sellers incur substantial per-consumer fixed costs and liquidity-
constrained consumers cannot afford to pay upfront fees equal to these fixed
costs. However, in the cell phone market, while fixed costs are high, they are
also endogenous. Carriers invest up to $400 in acquiring each new cus-
tomer, but much of these customer-acquisition costs are attributed to the
free or subsidized phones that carriers offer. This raises a series of questions.
Why do carriers offer free phones and lock-in contracts? Why not charge
customers the full price of the phone to avoid the lock-in? How many
consumers cannot afford to pay for a phone up-front? For how many of
these liquidity-constrained consumers is the carrier the most efficient source
of credit? The rational-choice model can explain the presence of lock-in
clauses, but only in a subset of contracts.
The rational-choice explanation for complexity is straightforward: Con-
sumers have heterogeneous preferences, and the complexity and multi-
dimensionality of the cellular service offerings cater to these heterogeneous
preferences. But while this heterogeneity likely explains some of the
observed complexity in the cell phone market, it cannot fully account for
the staggering level of complexity exhibited by the long menus of multi-
dimensional contracts available to consumers. Even for the rational con-
sumer, acquiring and comparing information on the range of complex
products is a time-consuming and costly undertaking. At some point, the
costs exceed the benefit of finding the perfect plan. Comparison-shopping
is deterred, and the benefits of the variety and multidimensionality are left
unrealized. It seems that in the cell phone market, the optimal level of com-
plexity has been exceeded.
C. Welfare Costs
The design of cellular service contracts is best explained as a rational response
to the imperfect rationality of consumers. Consumer mistakes and providers’
responses to these mistakes hurt consumers and generate welfare costs. For
example, consumers who misperceive their future use patterns choose the
wrong three-part tariff; that is, they do not choose the plan that would minimize
their total costs. Extrapolating from the sample of 3,730 subscribers described
above, the total annual reduction in consumer surplus from the three-part tariff
structure exceeds $13.35 billion. Moreover, while the average annual harm per
consumer, $47.68, is small, this average masks potentially important distribu-
tional implications. The $13.35 billion harm is not evenly divided among the
250 million U.S. cell phone owners. Many of these subscribers choose the right
plan. Even among those who choose the wrong plan, there is substantial het-
erogeneity in the magnitude of their mistakes. Each year, 42.5 million consum-
ers make mistakes that cost them at least 20 percent of their total yearly wireless
bill, or $146 per consumer annually. The distribution of mistakes implies a
potentially troubling form of regressive redistribution, since revenues from
consumers who make mistakes keep prices low for consumers who do not
make mistakes.
Other welfare costs are a consequence of lock-in. Lock-in prevents effi-
cient switching and thus hurts consumers. Switching is efficient when a dif-
ferent carrier or plan provides a better fit for the consumer. One survey found
that while 47 percent of subscribers would like to switch plans, only 3 percent
do so. The rest are deterred by the ETFs. Lock-in can also slow the beneficial
effects of consumer learning and prolong the costs of consumer mistakes,
since even consumers who learn from experience cannot benefit from their
new-found knowledge by immediately switching to another carrier’s plan.
(Insofar as carriers allow consumers to switch among their own monthly
plans, consumers can benefit from learning.) In addition to these direct costs,
lock-in may inhibit competition, adding a potentially large indirect welfare
cost. Since lock-in may prevent a more efficient carrier from attracting con-
sumers who are locked into a contract with a less efficient carrier, it can deter
new carriers from entering the market.2
Complexity is another detriment to welfare. The high level of complex-
ity of cell phone contracts can reduce welfare in two ways. First, consumers
tend to make more mistakes in plan choice when the menus are complex,
and these mistakes reduce consumer welfare. Second, complexity inhibits
competition by discouraging comparison-shopping. By raising the cost of
comparison-shopping, complex contracts reduce the likelihood that a con-
sumer will find it beneficial to carefully consider all available options.With-
out the discipline that comparison-shopping enforces, cellular service
2. A carrier’s relative efficiency depends on its costs of providing service and the quality of service
that it offers. Thus, a carrier that provides the same quality of service at lower cost than another
or a higher quality service at the same cost as another is a more efficient carrier.
c e l l phone s 191
providers can behave like quasi-monopolists, raising prices and reducing

consumer surplus.
D. Market Solutions and Their Limits

Do these behavioral market failures result from imperfect competition in
the cell phone market? The simple answer is “no.” In fact, enhanced com-
petition would likely make the identified design features more pervasive
and the resulting welfare costs higher. If consumers misperceive their future
use levels, competition will force carriers to offer three-part tariffs. If con-
sumers are myopic, competition will force carriers to offer free phones and
cover the cost of the subsidy with lock-in contracts. Finally, if consumers
ignore less salient price dimensions of complex, multidimensional contracts,
competition will force carriers to shift costs to these less salient price dimen-
sions. When demand for cellular service is driven by imperfect rationality,
competitors must respond to this biased demand; otherwise, they will lose
business and be forced out of the market. Accordingly, given consumers’
imperfect rationality, ensuring robust competition in the cellular service
market would not in itself solve the problem.
But it is a mistake to take the level of imperfect rationality as given. As we
have seen in previous chapters, competition, coupled with consumer learn-
ing, can reduce levels of bias and misperception and thus trigger a shift to
more efficient contractual design. In fact, the cellular service market has
exhibited numerous examples of such market correction in recent years and
now boasts a large set of products and contracts that cater to more sophisti-
cated consumers.
At the same time, however, the evolution of the market demonstrates
limits on the power of consumer learning to correct behavioral market
failures. For example, the market has responded to greater consumer aware-
ness of the costs of underestimated use among consumers who have expe-
rienced the sting of large overage charges. Since 2008, the major carriers
have been offering unlimited calling plans that arguably respond to demand
generated by this heightened consumer awareness. Yet, while overage fees
make it easy to learn the cost of underestimated use, the costs of overesti-
mated use are more difficult to learn since they are not so obviously penal-
ized. The result of this uneven learning is unlimited plans rather than the
optimal two-part tariff pricing scheme comprised of a fixed monthly fee
and a constant per-minute charge.
Another example: The shift from a time-invariant ETF to a time-variant,

graduated ETF structure responds to consumers’ increased awareness and
sensitivity to ETFs. This shift is not a pure market solution. Rather, it is an
example of how consumer learning and legal intervention can work in
tandem to change business practices. The change in ETF structure likely
began with a small number of consumers who learned to appreciate the
cost of ETFs and initiated litigation against the carriers.The threat of liabil-
ity and greater consumer awareness of ETFs then pushed carriers to adjust
their ETF structures.
Innovations like these suggest that the market has an impressive capacity
to correct for consumer misperceptions.Yet, market solutions are imperfect.
Not all biases are easily purged by learning. Not all consumers learn equally
fast, as evidenced by the limited adoption of many design innovations. The
speed of consumer learning and the market’s response matter, since welfare
costs are incurred in the interim period. Moreover, when consumers learn
to overcome one mistake, or when a previously hidden term becomes sali-
ent, carriers have an incentive to trigger a new kind of mistake or to add a
new non-salient term. Even if consumers always catch up eventually, this
cat-and-mouse game imposes welfare costs on consumers.
E. Policy Implications
While market solutions are imperfect and welfare costs remain, the poten-
tial for self-correction in the cellular service market merits a regulatory
stance that facilitates rather than impedes market forces; disclosure regula-
tion. The proposal we’ll explore deviates from existing disclosure regula-
tion and from most other proposals for heightened disclosure regulation.
Current disclosure regulation and other proposals focus on the disclosure
of product-attribute information; namely, information on the different fea-
tures and price dimensions of cellular service. The proposal we’ll explore,
by contrast, emphasizes the disclosure of use-pattern information, which as
you’ll recall from earlier chapters is information on how the consumer will
use the product.To fully appreciate the benefits and costs of a cellular serv-
ice contract, consumers must combine product-attribute information with
use-pattern information. For example, to assess the costs of overage fees, it
is not enough to know the per-minute charges for minutes not included
in the plan, as proposed in the Cell Phone User Bill of Rights. Consumers
must also know the probability that they will exceed the plan limit and by how
c e l l phone s 193
much. Use-pattern information can be as important as product-attribute

information. The disclosure regime should be redesigned to ensure that
consumers have access to both.
There are two possible approaches to disclosure regulation, approaches
that are not mutually exclusive. The first approach focuses on designing
simple disclosures that can be easily understood and utilized by imperfectly
rational consumers. In particular, carriers should provide total-cost-of-own-
ership (TCO) information, which is the total amount paid by the consumer,
given the consumer’s specific use patterns. This information should be
provided as an annual disclosure, such as on the year-end summary, to
account for month-to-month variations in use. The TCO disclosure com-
bines product-attribute information (pricing information) with informa-
tion on the specific consumer’s use patterns.This disclosure could be further
supplemented by information on alternative service plans that would reduce
the total price paid by consumers given their current use patterns.
The second approach to disclosure regulation re-conceptualizes disclosure,
targeting the disclosed information not directly at consumers but rather at
sophisticated intermediaries. Under this approach, carriers would provide
comprehensive, individualized use information in electronic, database form.
Imperfectly rational consumers will not try to analyze this information on
their own. Instead, they will forward the information to sophisticated inter-
mediaries. By combining the use information with the attribute informa-
tion they collect on product offerings across the cell phone market, the
intermediaries would be able to help each consumer find the plan that best
suits his or her specific use patterns.
The remainder of this chapter is organized as follows:
• Part I provides background information on the cell phone and the cel-
lular service market.
• Part II describes the key features of common cellular service contracts.
• Part III develops the behavioral-economics theory that explains these
contractual design features, after concluding that rational-choice explan-
ations fall short.
• Part IV discusses welfare implications.
• Part V considers the efficacy of market solutions.
• Part VI turns to policy, offering guidelines for enhanced disclosure
regulation.
I. The Cell Phone and the Cellular

Service Market
A. The Rise of the Cell Phone

1. Technology
The key technological innovation that underpins cellular communications
is the cellular concept itself. A cellular system divides each geographic mar-
ket into numerous small cells, each of which is served by a single, low-
powered transmitter. This allows the system to reuse the same channel or
frequency in non-adjacent cells in order to avoid interference. Thus, multi-
ple users can simultaneously make use of the same frequency. Sophisticated
technology locates subscribers and sends incoming calls to the appropriate
cell sites, while complex handoff technologies allow mobile consumers to
move seamlessly between cells.3
High demand for cellular service has prompted the development of
digital technology, which generates enhanced capacity without degrading
service quality. Two kinds of capacity-increasing technological solutions
have emerged. The first employs time-slicing technology; signals associated
with several different calls are aggregated within the same frequency by
assigning to each user a cyclically repeating time slot in which only that user
is allowed to transmit or receive. Time-slicing techniques include Bell Labs’
time division multiple access (TDMA) and Global System for Mobile
(GSM), which are used by AT&T and T-Mobile, and Integrated Digital
Enhanced Network (iDEN), which is used by Nextel. Spread spectrum
techniques, by contrast, spread many calls over many different frequencies
while using highly sophisticated devices to identify which signals belong
to which calls and decode them for end users. The family of digital
3. The information presented here, and below, on cellular technology and on the history of the
cellphone market is based, in large part, on: FCC, FCC 06–142, “Annual Report and Analysis
of Competitive Market Conditions with Respect to Commercial Mobile Services, Eleventh
Report” (2006) 21 F.C.C.R. 10947, 62 (hereinafter “FCC Eleventh Report”); Jonathan E.
Nuechterlein and Philip J. Weiser, Digital Crossroads (MIT Press, 2005); Mischa Schwartz, Mobile
Wireless Communications (Cambridge University Press, 2005); William Stallings, Wireless Com-
munications and Networking (Prentice Hall, 2002); SRI International,“The Role of NSF’s Support
of Engineering in Enabling Technological Innovation, Final Report Phase II 94–97” (1998),
<http://www.sri.com/policy/csted/reports/sandt/techin2/contents.html> (hereinafter “SRI-
NSF Report”); Theodore Rappaport, Wireless Communications (Prentice Hall, 1996).
c e l l phone s 195
standards employing spread spectrum technology is known as Code Division

Multiple Access (CDMA). CDMA standards are used by Verizon and Sprint.
The introduction of these digital cellular technologies, starting in the early
1990s, marked the advance from first-generation (1G) systems to second-
generation (2G) systems. Third-generation (3G) systems, which began to
operate in the U.S. in 2002, incorporate more advanced technologies that
provide the increased speed and capacity necessary for multimedia, data, and
video transmission, in addition to voice communications. And now fourth-
generation (4G) systems are being deployed.
2. History
Although the key concepts essential to modern cellular systems were con-
ceived in 1947, the Federal Communications Commission’s (FCC) refusal
to allocate substantial frequencies to mobile radio service meant that signifi-
cant development of cellular telephone services was delayed for several dec-
ades. It was not until the early 1980s that the FCC allocated 50 MHz of
spectrum in the 800 MHz band to cellular telephone service. The FCC
rules created a duopoly of two competing cellular systems in each of 734
“cellular market areas”—one owned by a non-wireline company and one
owned by the local wireline monopolist in the area. Each carrier received
25 MHz of spectrum. The first set of cellular licenses, which pertained to
the thirty largest urban markets (the Metropolitan Service Areas or MSAs)
were allocated by comparative hearings. However, the FCC was so over-
whelmed by the number of applicants that in 1984 Congress authorized the
use of a lottery system to allocate spectrum in the remaining markets. By
1986, all the MSA licenses had been allocated, and by 1991 licenses had been
allocated in all markets. As demand for cellular service rapidly increased over
subsequent years, the FCC allocated more spectrum to wireless communi-
cations. New spectrum has been allocated by auction rather than lottery
ever since Congress gave the FCC authority to issue licenses through auc-
tions in the 1993 Budget Act, a move designed to raise revenues and cut
down on delays associated with the lottery system.4
The more recent history of the cellular service market in the U.S. is
one of consolidation. As noted above, the industry began with the local
structural duopolies that were created by the FCC’s lottery mechanism.
4. See Omnibus Budget Reconciliation Act of 1993, Pub. L. No. 103-66, Title VI, § 6002(a),
6002(b)(2), 197 Stat. 312, 387–93 (1993) (codified as 47 U.S.C. § 309(j) (2006)).
With different firms operating in different geographical markets, the

national market initially included a large number of players. The number
of firms increased further as the FCC auctioned off more and more radio
spectrum for cell phone use. But this high level of market dispersion did
not last long. The FCC placed few restrictions on the ability of firms to
merge across markets, and a long history of voluntary merger and acquisi-
tion activity followed. Soon a handful of firms—AT&T Wireless, Cingu-
lar, Nextel, Sprint, T-Mobile, and Verizon Wireless—gained a dominant
position as nationwide carriers. Consolidation activity increased in 1999,
as national carriers sought to fill in gaps in their coverage areas and increase
the capacity of their networks while regional carriers sought to enhance
their ability to compete with the nationwide operators. Consolidation
was further facilitated by the FCC’s 2003 decision to abolish the regula-
tory spectrum cap that had limited the amount of spectrum that a com-
pany could own in any one geographical market, since this opened the
door to mergers by companies with overlapping coverage areas. Most sig-
nificantly, in October 2004, Cingular and AT&T Wireless merged to
become AT&T Wireless, while in December 2004 Sprint and Nextel
merged to become Sprint Nextel.5
3. Economic Significance
The FCC estimates that at the end of 2009, there were 291 million cellular
service subscribers in the U.S., which corresponds to a nationwide
penetration rate of 93 percent. The market has been growing rapidly, albeit
with signs that the market is approaching saturation. Cellular service pro-
viders added 11.1 million new subscribers in 2009, 16.6 million in 2008,
21.2 million in 2007, 28.8 million in 2006, 28.3 million in 2005, 24.1 mil-
lion in 2004, and 18.8 million in 2003. An historical perspective under-
scores the stellar growth of the market; 286 million subscribers were added
between June 1990 and the end of 2009. While cell phones complement
landline phones for most users, a significant and increasing number of users
view the cell phone as a partial or even complete replacement for the
traditional, landline phone. In the first half of 2010, an estimated 26.6 percent
5. FCC, FCC 05–173, “Annual Report and Analysis of Competitive Market Conditions with
Respect to Commercial Mobile Services, Tenth Report” (2005) 20 F.C.C.R. 15908, 15930 ¶ 58
(2005); Jeremy T. Fox, “Consolidation in the Wireless Phone Industry” (2005) 3, 7, 9 (Net Inst.
Working Paper No. 05-13), available at <http://www.netinst.org/Fox2005.pdf>.
c e l l phone s 197
of households used only wireless phones, up from 4.2 percent at the end of
2003.6
The high revenues enjoyed by carriers provide an indication of the
magnitude of the cellular service market. In the second quarter of 2011,
Verizon posted wireless revenues of $17.3 billion, AT&T $15.6 billion,
Sprint $7.5 billion, and T-Mobile $5.1 billion.7 Total quarterly wireless
revenues for the four national carriers were $45.5 billion, which poten-
tially translates into total annual wireless revenues of $182 billion, ignoring
seasonal variations. Wireless telecommunications have become the largest
source of profit for nearly all major telecommunication providers. For
example, Verizon’s wireless services are about twice as profitable as its
wireline offerings.8 Looking at revenues from spectrum auctions is also
instructive. In 2006, the FCC’s Auction No. 66 raised a total of $13.7 bil-
lion in net bids from wireless providers for 1,087 spectrum licenses in the
1710–1755 MHz and 2110–2155 MHz bands.9 In 2008, the FCC’s Auction
No. 73 raised a total of $19.0 billion in net bids from wireless providers for
1,099 licenses in the 698–806 MHz band (known as the “700 MHz
Band”).10
Investment in telecommunications infrastructure in general—and one
could argue cellular technology in particular—promotes economic growth
Respect to Commercial Mobile Services, Fifteenth Report” (2001) 8, 9, 207, available at
<http://wireless.fcc.gov/index.htm?job=cmrs_reports> (hereinafter “FCC Fifteenth Report”;
FCC, DA 09-54, “Annual Report and Analysis of Competitive Market Conditions with
Respect to Commercial Mobile Services, Thirteenth Report” (2009) 24 F.C.C.R. 6185, at
6279–80 ¶ 197, 6301 ¶ 230 (hereinafter “FCC Thirteenth Report”); SRI-NSF Report, above
note 3, at 94.
7. Verizon, “Verizon Wireless—Selected Financial Results” ( July 22, 2011), <http://www22
.verizon.com/idc/groups/public/documents/adacct/2011_2q_fs_pdf.pdf>; AT&T, “Investor
Briefing 2nd Quarter 2011,” at 4 ( July 21, 2011), <http://www.att.com/Investor/Financial/
Earning_Info/docs/2Q_11_IB_FINAL.pdf>; News Release, “Sprint, Sprint Nextel Reports
Second Quarter 2011 Results” ( July 28, 2011), <http://newsroom.sprint.com/article_display
.cfm?article_id=1990>; Financial Release, “T-Mobile USA,” T-Mobile USA Reports Second
Quarter 2011 Results” Bibliog cites July 28, 2011, <http://www.t-mobile.com/company/
Investor Relations.aspx?tp=Abt_Tab_InvestorRelationsandViewArchive=Yes> (follow
“T-Mobile USA Reports Second Quarter 2011 Results” hyperlink).
8. George Gilder, “The Wireless Wars,” Wall Street Journal, April 13, 2007, at A13 (stating that
Verizon’s mobile phones generated $804 million in profits, whereas its wired phones gener-
ated $393 million in profits).
9. “Auction of Advanced Wireless Services Licenses Closes: Winning Bidders Announced for
Auction No. 66” (2006) 21 F.C.C.R. 10521.
10. “Auction of 700 MHz Band Licenses Closes: Winning Bidders Announced for Auction 73”
(2008) 23 F.C.C.R. 4572.
by reducing the costs of interaction, expanding market boundaries, and

enhancing information flow. Specifically, cellular technology can create
value by facilitating communication between individuals who are on the
move, thus helping individuals to better coordinate their activities and
respond to unforeseen contingencies.Wireless services also boost growth by
expanding telephone networks to include previously disenfranchised con-
sumers through prepaid service that is unavailable for fixed lines. Analysts
estimate that the decades-long delay in the development of cellular net-
works after the discovery of the cellular concept cost the U.S. economy
around $86 billion (measured in 1990 dollars).11
B. The Cellular Service Market

1. Structure
The U.S. cellular service industry is dominated by four “nationwide” facil-
ities-based carriers: AT&T Wireless, Verizon Wireless, Sprint Nextel, and
T-Mobile. At the end of 2010, each had networks covering at least 250 mil-
lion people. AT&T had 95.5 million subscribers,Verizon 94.1 million, Sprint
Nextel 49.9 million, and T-Mobile 33.7 million.12
In addition to the national carriers, there are a number of regional carri-
ers, including Leap, U.S. Cellular, and MetroPCS. There is also a growing
resale sector, consisting of providers who purchase airtime from facilities-
based carriers and resell service to the public, typically in the form of pre-
paid plans rather than standard monthly tariffs.
11. See Robert Jensen, “The Digital Provide: Information (Technology), Market Performance,
and Welfare in the South Indian Fisheries Sector”, Q.J. Econ., 122 (2007), 879, 881–3 (describ-
ing how the introduction of cell phones revolutionized the fishing industry in Kerala, lead-
ing to dramatic reductions in price dispersion, the complete elimination of waste (previously
5–8 percent of the daily catch), an 8 percent average increase in fishermen’s profits, a 4 per-
cent decline in consumer prices, and a 6 percent increase in consumer surplus); Leonard
Waverman, Meloria Meschi, and Melvyn Fuss, “The Impact of Telecoms on Economic
Growth in Developing Countries,” in The Vodafone Policy Paper Series no. 3, Africa: The Impact
of Mobile Phones (2005), <http://www.vodafone.com/etc/medialib/attachments/cr_
downloads.Par.78351.File.tmp/GPP_SIM_paper_3.pdf> (“We find that mobile telephony
has a positive and significant impact on economic growth, and this impact may be twice as
large in developing countries as compared to developed countries.”); Nuechterlein and
Weiser, above note 3, at 268.
12. Information presented here, and below, on the cellular service market is based, in large part,
on FCC Fifteenth Report, above note 6.
c e l l phone s 199
2. Competition
The overlapping geographic coverage of the national and regional providers
gives rise to competition between cellular service providers. The FCC esti-
mates that 97.2 percent of people have three or more different operators offer-
ing cell phone services in the census blocks where they live, 94.3 percent live
in census blocks with four or more operators, 89.6 percent live in census blocks
with five or more operators, 76.4 percent live in census blocks with six or more
operators, and 27.1 percent live in census blocks with seven or more operators.
The FCC measures market concentration by computing the average Herfind-
ahl-Hirschman Index (HHI) across 172 Economic Areas (EAs)—aggregations
of counties that have been designed to capture the “area in which the average
person shops for and purchases a mobile phone, most of the time.” The HHI
is a measure of market concentration that ranges from a value of 10,000 in a
monopolistic market to zero in a perfectly competitive market.13 In mid-2010,
the average HHI, weighted by EA population, was equal to 2,848. An industry
with an HHI above 2,500 is considered highly concentrated by the antitrust
authorities. And these figures might well underestimate market concentration,
since the FCC’s methodology gives equal weight to a mobile carrier assigning
cell phone numbers in one county as it does to a carrier that assigns numbers
in multiple counties in a given EA.14 Indeed, one analyst calculated an average
HHI value exceeding 6,000 with 2005 data, using the amount of spectrum
controlled by a carrier in a market as a proxy for market share.15
The relatively high level of concentration in the cell phone market is the
product of an ongoing consolidation process. This consolidation is at least
partly motivated by a desire to realize economies of scale and enlarge geo-
graphic scope. Broad coverage can be provided at lower cost by a single
nationwide carrier than by regional carriers through roaming agreements
with carriers operating in different geographic areas. In addition, extending
I
13. Formally, the HHI is given by HHI = ∑ (100 si ) , where si is the fractional market share of
2
i =1
firm i, and I is the number of firms in the market. Thus a monopolistic market has an HHI of
10,000, a market that is equally divided between two firms has an HHI of 5,000, a market that
is equally divided between three firms has an HHI of 3,333.33, a market that is equally divided
between four firms has an HHI of 2,500, etc.
14. FCC Thirteenth Report, above note 6, at 6212 ¶ 45 n. 87.
15. Fox, above note 5, at 15–17. Moreover, this figure excludes data on Nextel, and so the Sprint
Nextel merger does not contribute to the high HHI, suggesting that this figure may underes-
timate the true concentration. Id. at 16 n. 11.
the national network spreads fixed costs, such as marketing expenditures

and investments in developing new technology, over a wider base of
customers. Economies of geographic scope arising from complementarities
between markets may also provide an efficiency reason for consolidation.16
However, even if consolidation reduces certain costs, other costs may
increase. Consolidation tends to reduce competition and facilitate collusion
as the number of multi-market contacts between the dominant national
carriers increases.17
The magnitude of entry barriers provides another important measure of
competitiveness. If entry barriers are low, even a market with a small number
of firms will behave competitively. Government control of spectrum—limit-
ing the amount of spectrum allocated to wireless communications and requir-
ing carriers to obtain a government-issued license—has the potential to create
significant barriers to entry. However, the FCC has alleviated many of these
concerns recently by increasing the amount of spectrum available for cellular
communication services and allowing market forces to determine market
structure through elimination of the old structural duopolies and abolition of
the spectrum cap. Moreover, the Telecommunications Act and FCC regula-
tions reduce entry barriers by imposing interconnection and roaming obliga-
tions. The ability to purchase spectrum on the secondary market further
reduces entry barriers.18 Meanwhile, advertising expenditures—amounting to
billions of dollars annually19—and the economies of scale and scope described
above continue to impose substantial entry barriers.
16. See Patrick Bajari, Jeremy T. Fox, and Stephen Ryan, “Evaluating Wireless Carrier Consolidation
Using Semiparametric Demand Estimation” (2006) 5 (Nat’l Bureau of Econ. Research,Working Paper
No. 12425), available at <http://www.nber.org/papers/w12425>; Fox, above note 5, at 10.
17. Fox, above note 5, at 12. Multi-market contact was an important factor in explaining above-
competitive prices in the early mobile telecommunications industry. See Philip M. Parker and
Lars-Hendrik Röller, “Collusive Conduct in Duopolies: Multi-Market Contact and Cross-
Ownership in the Mobile Telephone Industry”, Rand J. Econ., 29 (1997), 304, 320.There were
also significant cross-ownership effects, i.e., if operators co-own an operating license else-
where, they tend to collude more. Id.
18. On the FCC actions—making more spectrum available, eliminating the structural duopolies
and abolishing the spectrum cap—see FCC Thirteenth Report, above note 6, at 6220 ¶¶ 65–6.
On interconnection and roaming obligations—see 47 U.S.C. § 251(a)(1) (2006); “Reexamina-
tion of Roaming Obligations of Commercial Mobile Radio Service Providers, Final Rule”
(2007) 72 Fed. Reg. 50064, 50064–65 (hereinafter “Reexamination of Roaming Obligations”).
On the role of the secondary market—see FCC Thirteenth Report, above note 6, at 6220 ¶
67.
19. Estimated advertising spending for wireless telephone services totaled between $4.1 billion
and approximately $5.1 billion in 2007, $3.7 billion in 2008, and $3.4 billion in 2009. See FCC
Thirteenth Report, above note 6, at 6261 ¶ 158 (for the 2007 estimates); FCC Fifteenth
Report, above note 6, at 50 (for the 2008 and 2009 estimates).
c e l l phone s 201
Switching costs also affect the level of competition. Switching costs in

the cellular service market are substantial, although recent developments are
reducing these costs. Until recently, most consumers signed long-term con-
tracts with fixed ETFs of approximately $200. But now major carriers are
offering contracts with graduated ETFs that decline over the life of the
contract. Likewise, historically carriers have allowed only certain approved
phones to be used by their subscribers on their network and “locked” the
phones they sold to render them incapable of being used on other net-
works.20 The recent trend, however, is toward open access, which allows
more phones onto the network, and recent regulatory action by the Copy-
right Office clarified that phones can be unlocked.21 Being forced to change
phone numbers was also a potentially significant switching cost until it was
eliminated by the regulatory requirement that carriers provide local number
portability.22 The high churn rates in the cell phone market—between 1.5
percent and 3.3 percent per month in 200923—suggest that switching costs,
while potentially substantial, are not prohibitive for many consumers.
To sum up, while there is reason to believe that the cellular service mar-
ket is less than perfectly competitive, cellular service providers are actively
competing to attract consumers. Declining prices, albeit with a leveling off
in recent years, are evidence of such active competition.24 Competition is
20. Tim Wu,“Wireless Net Neutrality: Cellular Carterfone and Consumer Choice in Mobile Broad-
band” (2007) 1 New Am. Found.Wireless Future Program Working Paper No. 17, available at <http://
www.newamerica.net/files/WorkingPaper17_WirelessNetNeutrality_Wu.pdf>; see also Spencer
E. Ante, “Verizon Embraces Google’s Android”, Business Week, December 3, 2007, <http://www.
businessweek.com/technology/content/dec2007/tc2007123_429930.htm?campaign_id=yhoo>
(“Verizon Wireless has created the most profitable U.S. cellular business by tightly restricting the
devices and applications allowed to run on its network.”).
21. See 37 C.F.R. § 201.40(b)(5) (2008). Carriers are embracing the new open-access business
model. See Ante, above note 20. (“But over the past year, [Verizon’s] leadership came to con-
clude that it was time for a radical shift. Such a move, they reckoned, might help Verizon
Wireless keep growing while holding down costs.”) Sprint Nextel and T-Mobile also support
the shift to an open-handset environment, as members of the Google-led “Open Handset
Alliance.” Id.; see also Amol Sharma and Dionne Searcey, “Verizon to Open Cell Network to
Others’ Phones,” Wall Street Journal, November 28, 2007, at B1.
22. FCC Eleventh Report, above note 13, at 11012 ¶ 146. Wireless local number portability began
on November 24, 2003. In re Telephone Number Portability, 19 F.C.C.R. 875, 876 (2) (order).The
FCC reports that from December 2003 to December 2007, 49.93 million consumers took
advantage of the right to retain their phone number while switching from one wireless carrier
to another. FCC Thirteenth Report, above note 6, at 6272 ¶ 183.
23. FCC Fifteenth Report, above note 6, at 154–5. A “churn rate” is the rate at which users cancel
their cellular service in a given period of time.
24. FCC Fifteenth Report, above note 6, at 12–13. On the other hand, there is substantial similar-
ity between the pricing schemes offered by the major carriers. See below Part II. This price
matching may reflect tacit collusion among the major carriers. Cf. Meghan R. Busse, “Multi-
market Contact and Price Coordination in the Cellular Telephone Industry” (2008–9) 9
J. Econ. and Mgmt. Strategy, 9 (2008-9) 287, 313–16.
also observed on non-price dimensions. Competition to attract and retain

customers appears to be driving carriers to improve service quality. Carriers
pursue a variety of strategies to improve service quality, including network
investment to improve coverage and quality and acquisition of additional
spectrum.25 While an economic conclusion reached by politically appointed
regulators should be taken with a grain of salt, it is noteworthy that the
FCC described the cellular service market as one characterized by healthy
competition with carriers engaging in “independent pricing behavior, in
the form of continued experimentation with varying pricing levels and
structures, for varying service packages, with various handsets and policies
on handset pricing.”26
3. Related Markets
The cellular service market interacts with other markets, specifically with
the market for phones/handsets and with the market for cell phone
applications.
a. The Handset Market

The market for handsets is controlled by five firms: Samsung, LG Electron-
ics, Motorola, Apple, and RIM. In the U.S., Samsung enjoys the largest
market share, controlling 25.5 percent of the handset market in the second
quarter of 2011. LG placed second with 20.9 percent of the market, and
Motorola followed with 14.1 percent. Apple and RIM, the maker of Black-
Berry, lag behind considerably with 9.5 percent and 7.6 percent of the
market, respectively.27
In the U.S., the major cellular service providers exert significant con-
trol over the handset market. Internationally, about half of handsets are
25. FCC Thirteenth Report, above note 6, at 6262–3. Carriers’ marketing campaigns emphasize
their “superior network coverage, reliability, and voice quality.” Id.
26. FCC Eleventh Report, above note 3, at 10987 ¶ 90.Yet, since this is an industry characterized
by high network costs, this phase of apparently intense competition may be nothing more
than a price war designed to squeeze out smaller carriers that will ultimately result in an
increase in the market power of the remaining large carriers and an attendant rise in prices.
27. Press Release, comScore, “comScore Reports July 2011 U.S. Mobile Subscriber Market Share”
(August 30, 2011), <http://www.comscore.com/Press_Events/Press_Releases/2011/8/
comScore_Reports_ July_2011_U.S._Mobile_Subscriber_Market_Share>. The market shares
of the leading handset firms are quite different outside the United States. Nokia is the global
market leader, with 33.3 percent of the global market in 2010, followed by Samsung with 20.6
percent, LG with 8.6 percent, RIM with 3.6 percent, and Apple with 3.5 percent. Press
Release, “Strategy Analytics, Global Handset Shipments Reach 400 Million Units in Q4
2010” ( January 28, 2011), <http://www.strategyanalytics.com/default.aspx?mod=pressrelease
vieweranda0=5001>.
c e l l phone s 203
purchased through carriers and about half are sold directly to consumers
through other channels. In the U.S., the vast majority of cell phones—
nine out of every ten phones according to one estimate—are sold through
a service provider.28 The practice of subsidizing handset prices for con-
sumers who sign long-term service contracts is at least partially responsi-
ble for the competitive disadvantage suffered by handset makers looking
to sell directly to consumers.
Carriers in the U.S. determine which devices consumers can operate on
their networks. The result of this control by service providers is that only a
fraction of any given manufacturer’s total line of products is offered. For
example, in 2006, of the fifty new products Nokia introduced into the market,
U.S. cellular service providers offered a scant few. By allowing only certain
approved phones on their networks, carriers influence the design of handsets.
Moreover, as a condition of network access, carriers require that developers
disable certain services or features that might be useful to consumers, such as
call-timers, photo sharing, Bluetooth capabilities, and Wi-Fi capabilities.29
But the balance of power is shifting. Handset brands and models are an
increasingly important determinant of a consumer’s choice of service pro-
vider. Apple’s launch of the iPhone is a significant example of a handset
manufacturer successfully overcoming carrier pressure. More generally, the
rapid expansion of the “smartphone” market is enhancing the power of
handset makers and companies who provide software for these handsets. An
example is the Android operating system, developed by Google.30
In addition, the open-access trend is starting to limit carriers’ control
over the handset market.31 Regulation is also playing an important role:
28. Marguerite Reardon, “Unlocking the Unlocked Cell Phone Market,” CNET News, July 2,
2009, <http://news.cnet.com/8301-1035_3-10277723-94.html>.
29. See Wu, above note 20, at 10–13; Reardon, above note 28; Ante, above note 20 (“Verizon
Wireless has created the most profitable U.S. cellular business by tightly restricting the devices
and applications allowed to run on its network.”).
30. On the increasing power of handset manufacturers—see Rita Chang, “Proof That Handset
Brands Help Sell Wireless Plans,” RCR Wireless, October 28, 2008, <http://www.rcrwireless
.com/article/20081028/WIRELESS/810289995/1081/proof-that-handset-brands-help-sell
-wireless-plans#>; Press Release, Nielsen, “In US, Smartphones Now Majority of New Cell-
phone Purchases” ( June 30, 2011), <http://blog.nielsen.com/nielsenwire/?p=28237>. On
power struggles between carriers and handset manufacturers, as well as with application devel-
opers—see generally Jessica E. Vascellaro, “Air War: A Fight over What You Can Do on a
Cellphone,” Wall Street Journal, June 14, 2007, at A1; see also Miguel Helft and Stephen Laba-
ton, “Google Pushes for Rules to Aid Wireless Plans,” New York Times, July 21, 2007, at A1.
31. See George S. Ford, Thomas M. Koutsky, and Lawrence J. Spiwak, “Wireless Net Neutrality:
From Carterfone to Cable Boxes”, Phoenix Ctr. Pol’y Bull. 17, (April 2, 2007), at 2, <http://
phoenix-center.org/PolicyBulletin/PCPB21Final.pdf>.
One-third of the recently auctioned spectrum comes with a requirement

that “cellular networks allow customers to use any phone they want on
whatever network they prefer, and be able to run on it any software they
want.”32 And, perhaps sensing the inevitable, carriers are beginning to
embrace the new open-access business model, reasoning that they can cut
costs by eliminating handset subsidies and letting handset manufacturers
bear most of the development and customer service costs.33
b. The Applications Market

The major cellular service providers and other mobile data providers have
progressively introduced a wide variety of mobile data services and applica-
tions, including text and multimedia messaging, ringtones, GPS navigation,
and entertainment applications from games to TV and music players.34 Data
revenues have been growing—in absolute numbers and as a share of total
revenues. In 2009, $42 billion or 27 percent of total wireless service revenues
were from data revenues.35
The major carriers exert substantial control over the applications market.
Many applications—popularly known as “apps”—are often sold by the car-
riers as part of the service package. Although some application developers
sell their applications directly to consumers, carriers exert considerable
influence over the design, content, and pricing of cell phone applications.
For example, carriers impose limits on “unlimited use” pricing plans for 3G
broadband data services by restricting bandwidth and designating certain
applications as “forbidden” in consumer contracts. Carriers also create
obstacles for application developers by restricting access to many phone
capabilities, imposing extensive qualification and approval requirements
before allowing them to develop applications for their cell phone platforms,
and by failing to develop uniform standards.36
As sophisticated new applications for cell phones have proliferated, how-
ever, handset manufacturers have started to put pressure on carriers to loosen
their grip on the applications market. For example, the immense popularity
32. Editorial, “A Half-Win for Cellphone Users,” New York Times, August 6, 2007, at A18; In re
Service Rules for the 698–746, 747–762, and 777–792 MHz Bands, 22 F.C.C.R. 15289, 15367,
15370–71 (2007) (second report and order) (hereinafter “Service Rules Second Report and
Order”).
33. See Ante, above note 20; see also Sharma and Searcy, above note 21.
34. FCC Eleventh Report, above note 3, at 11007 ¶ 136–37.
35. FCC Fifteenth Report, above note 6, at 13–14.
36. Wu, above note 20, at 13–14, 22–5.
c e l l phone s 205
of the iPod music player allowed Apple to persuade AT&T to sell the iPhone
to its customers without also offering AT&T’s own line of applications.37
II. The Cellular Service Contract

Cellular service contracts are complex multidimensional contracts. This
chapter does not attempt a comprehensive analysis of these contracts.38
Rather, the focus will be on the three design features mentioned earlier: (1)
the three-part tariff structure, (2) the lock-in clause, and (3) the high level
of complexity itself.39
A. Three-Part Tariffs
As noted earlier, cellular service contracts are complex and multidimen-
sional. To begin with, most postpaid plans, which constitute the majority of
plans, price their basic voice-calling service using a three-part tariff struc-
ture. This structure is a three-dimensional pricing scheme that includes a
monthly charge, a number of included voice minutes, and a per-minute
price for minutes beyond the plan limit (“overage”). Higher-priced plans
with a higher monthly charge come with more allotted minutes and lower
overages for minutes exceeding the plan limit. For example, as of this writ-
ing, AT&T, Sprint, and Verizon offer a $39.99 plan with 450 minutes and
37. Vascellaro, above note 30.

38. One feature that we will not address is the definition of call types for which the subscriber is
charged (or that count toward the plan limit). Specifically, while in most countries subscribers
are charged only for outgoing calls, in the U.S. subscribers are also charged for incoming calls.
This feature of the U.S. cellular service market seems to fit nicely within the general behav-
ioral theory, as subscribers probably find it even more difficult to accurately estimate the
number/length of incoming calls along with outgoing calls than outgoing calls alone.
39. The description of products and prices provided in Part II, and in other Parts, is largely based
on information available through carriers’ websites focusing on services available in the New
York area. See AT&T, “Cell Phones, Cell Phone Plans, and Wireless Accessories—from
AT&T”, <http://www.att.com/shop/wireless/#fbid=n6la6zd-YLG> (last visited September
29, 2011); Sprint, “Cell Phones, Mobile Phones, and Wireless Calling Plans from Sprint,”
<http://www.sprint.com> (last visited September 29, 2011); T-Mobile, “Cell Phones | 4G
Cell Phone Plans | Android Tablet PCs | T-Mobile,” <http://www.t-mobile.com/> (last
visited September 29, 2011);Verizon Wireless, “Cell Phones—Smartphones: Cell Phone Serv-
ice, Accessories—Verizon Wireless,” <http://www.verizonwireless.com/b2c/index.html>
(last visited September 29, 2011). It should be noted that some variation exists between online
and offline (retail store) offerings and between different geographical markets across the U.S.
This variation is mentioned explicitly only when it is relevant to the analysis.
$0.45 per-minute overage, a $59.99 plan with 900 minutes and $0.40 per-
minute overage, and a $79.99 plan with 1350 minutes and $0.35 per-minute
overage.
The three-part tariff was introduced in the U.S. in 1998. Before then, all
wireless plans involved roaming and long-distance charges.40 In 1998, AT&T
began offering a plan that allowed customers to pay a fixed monthly fee for
a set number of minutes that could be used for both local and long-distance
calls. As a result, AT&T gained 850,000 customers in its first year, perhaps
more customers than it could serve.41 AT&T’s competitors soon followed
with similar pricing plans. Much of the rising usage of cellular service was
attributed to this pricing structure.42
Industry accounts of the reason for the switch to bundle pricing vary.
Some argue that bundle pricing responds to consumer demand for simpli-
city.43 Others, including AT&T’s CEO at the time, Mike Armstrong, suggest
that the move to bundle pricing was motivated by a desire to attract heavy
users. Armstrong’s account is consistent with two key facts: (1) the smallest
fixed fee offered was $90 per month, and (2) after the introduction of its
One Rate plan, the average AT&T subscriber bill increased, raising the
company’s profitability.44
B. Lock-In Clauses
In addition to the three-part tariff pricing structure, most postpaid call-
ing plans share the following two features; a free or substantially dis-
counted phone and long-term contracts with ETFs. At the time of this
writing, AT&T gave customers the option to buy the LG Phoenix for
$379.99 without a contract. With the signing of a two-year data contract,
however, AT&T charged $0.01 for the same phone. T-Mobile customers
40. See Elizabeth Douglass, “The Cutting Edge Special Report: Wireless Communications; ‘Pre-
paid’ Idea is Catching On in U.S. Market,” L.A.Times, March 15, 1999, at C1 (discussing trend
away from long-distance and roaming charges).
41. Roger O. Crockett, “The Last Monopolist,” Business Week, April 12, 1999, at 55; Dan Meyer,
“Coverage Problems Trigger Headaches for Carriers,” RCR Wireless News, July 9, 2001, at 16.
42. Andrew M. Odlyzko, “The Many Paradoxes of Broadband” First Monday, September 1, 2003,
<http://firstmonday.org/htbin/cgiwrap/bin/ojs/index.php/fm/article/view/1072/992>.
43. See Rebecca Blumenstein, “The Business-Package Plan: AT&T Sees Wireless as the Key to its
Broader Strategy of Bundling Its Services,” Wall Street Journal, September 20, 1999 at R26; see
also Peter Elstrom, “Wireless with All the Trimmings,” Business Week, November 16, 1998.
44. Peter Elstrom, “Mike Armstrong’s Strong Showing,” Business Week, January 25, 1999, at 94;
Elstrom, “Wireless with All the Trimmings,” above note 43.
c e l l phone s 207
who signed a new contract had the opportunity to receive, for free, a
number of phones with suggested retail values of over $200, including
the LG Optimus, the HTC Wildfire, and the Samsung Gravity. Similarly,
AT&T and Apple heavily subsidized the iPhone, sacrificing short-term
revenues, and Sprint sold Samsung’s music phones far below cost at only
$149.45 The free or heavily subsidized phone strategy pervades the U.S.
cell phone market. A 2011 survey by J.D. Power found that 42 percent of
customers receive a free cell phone when subscribing to a wireless
service.46
Of course, free phones are not really free. Carriers recoup the costs of the
phones through subscription fees.To make sure that they collect enough sub-
scription fees to cover the cost of the phone, carriers lock consumers into
long-term contracts.47 Such lock-in is secured by substantial ETFs. For exam-
ple, in June 2007, T-Mobile charged a fixed termination fee of $200, AT&T
charged $175, and Sprint charged up to $200, depending on the service
selected. Historically, the same termination fees were charged regardless of
when the agreement was broken, meaning that a consumer would have paid
a $200 termination fee for ending a two-year contract just one month early.
In the wake of a number of class action lawsuits challenging the legality of
these fees, providers have begun to offer contracts with termination fees that
decline over the life of the contract.Verizon led this transition when, in June
2007, it started charging customers a termination fee of $175, less $5 for each
full month that the customer remains on the initial contract. By the end of
2008, all the major carriers were offering similar graduated ETFs.48
45. See Amol Sharma and Roger Cheng, “iPhone Costs Prove a Drag for AT&T,” Wall Street Jour-
nal, October 23, 2008, at B4 (“The company said $900 million in customer-acquisition costs
related to the iPhone shaved 10 cents off its earnings,” but “AT&T executives said the invest-
ment will pay off because iPhone users are lucrative in the long-term, spending about $95 a
month on average, or about 1.6 times the amount other customers do.”); Cliff Edwards and
Roger O. Crockett, “New Music Phones—Without the i,” Business Week, April 16, 2007, at 39.
46. Press Release, J.D. Power and Associates, “The Right Blend of Design and Technology is
Critical to Creating an Exceptional User Experience with Smartphones and Traditional
Mobile Devices” 2 (September 8, 2011), <http://businesscenter.jdpower.com/JDPAContent/
CorpComm/News/content/Releases/pdf/2011146-whs2.pdf>.
47. When no-contract plans are offered, phone subsidies disappear. For example, a customer with
no contract would be required to pay an additional $400 beyond the contract price for the
same iPhone. “AT&T Plans to Offer No-Contract iPhone,” Wall St. J., July 2, 2008, at B5.
48. See Verizon Wireless, Customer Agreement, <http://www.verizonwireless.com/b2c/index
.html> (follow “Customer Agreement” hyperlink at the bottom of the page); AT&T, Plan
Terms, <http://www.wireless.att.com/cell-phone-service/legal/plan-terms.jsp#gsm>; Press
Release, Sprint Nextel Corp.,“Sprint Launches One of the Industry’s Most Customer-Friendly
C. Complexity
The complexity and multidimensionality of cellular service contracts can
be viewed as a contractual design feature. Most cellular service contracts are
highly complex in and of themselves.This high level of complexity increases
substantially when the perspective shifts from a single-contract to the many
different multidimensional contracts being offered. According to one indus-
try estimate, the cellular service market boasts “more than 10 million differ-
ent plans and add-on combinations.”49
1. Postpaid Plans—The Basics

Even the basic components of the common postpaid calling plan are com-
plex. As described above, the basic pricing scheme is three-dimensional.
Each dimension of the basic pricing scheme is one of the tariffs that make
up the three-part tariff. Moreover, each provider offers a long menu of dif-
ferent three-part tariffs. To make things even more complicated, the menus
of three-part tariffs vary among providers.50 Further complexity is intro-
duced by the diversity of additional service features covered by the fixed
monthly fee. Some of these features are offered by all carriers in the exact
same way. Others are offered by some carriers but not others or are offered
in varying formats by the different carriers.
For example, all four major carriers offer unlimited calls during off-peak
times, usually nights and weekends. There is, however, some potentially sig-
nificant variation. Nights are defined differently by different carriers. For
AT&T and Verizon, the night begins at 9 p.m. and ends at 6 a.m. For
T-Mobile, the night begins at 9 p.m. and ends at 7 a.m. For Sprint, the night
begins at 7 p.m. and ends at 7 a.m., except for its more basic plans, where
the night begins at 9 p.m. and ends at 7 a.m. By varying the definition of
“night,” providers can offer up to three extra hours of unlimited calling.
Policies on Pro-Rated Early Termination Fees” (October 31, 2008), <http://newsroom.sprint

.com/article_display.cfm?article_id=771>; T-Mobile, T-Mobile Terms and Conditions,
<http://www.t-mobile.com/templates/popup.aspx?passet=ftr_ftr_termsandconditions>.
49. See BillShrink.com, Cell Phone Plans, Compare Best Cellular Service Carrier Deals on
BillShrink, <http://www.billshrink.com/cell-phones/plans.html> (last visited September 29,
2011).
50. We briefly mention two additional dimensions: (1) The directionality of the calls that con-
sume allotted minutes, and (2) the one-time activation charge. Along dimension (1), allotted
minutes are typically used up on both outgoing calls and incoming calls. As for (2), AT&T and
Sprint charge a $36 activation fee while Verizon and T-Mobile charge $35.
c e l l phone s 209
These three extra hours represent an additional 33.3 percent of unlimited

calling time. But since most consumers probably talk more during the three
hours between 7 and 9 p.m. and between 6 and 7 a.m. than they do during
the three hours between, say 1 and 4 a.m., these extra three hours of unlim-
ited calling probably represent much more than a 33.3 percent increase in
value.
To take another example, consumers might also consider whether to
select Verizon’s Friends and Family program, offering unlimited calls to five
(or ten) phone numbers selected by the user. Sprint, T-Mobile, and AT&T
offered similar programs, but as of this writing their plans were no longer
available to new users.
2. Family Plans
We have thus far focused on individual calling plans.The four major carriers
also offer family plans, adding another layer of complexity. The identifying
feature of a family plan is the ability to share the allotted minutes between
up to five users, each operating on a different line. For example, Verizon
offers family plans with monthly charges ranging from $69.98 to $119.98,
allotted minutes ranging from 700 to unlimited, and overages ranging from
$0.45 to $0.35. These monthly prices include two phone lines, and families
can add up to three more lines for an additional $9.99 per month per line
(or $49.99 under the unlimited plan).
3. Add-Ons
Cell phones can be used for much more than voice communication.
Carriers offer advanced communication services, including text messag-
ing, multimedia messaging, and Internet and email data services. Carriers
also offer applications, such as ringtones and games, as well as monthly
mobile Internet access packages. These services and applications are mar-
keted to consumers primarily as add-ons to their voice services.
Pricing of these services adds additional complexity. Providers offer
advanced communication services to consumers in one of three modes:
(1) Pay-as-you-go, applied mainly to text and multimedia messaging, where
the consumer pays per message sent or received
(2) Fixed-quantity monthly packages, where the consumer pays a monthly
fee for a fixed number of allotted messages or megabytes of data
(3) Unlimited-quantity monthly packages, where the consumer pays a

monthly fee for unlimited messaging or data transmission.51
Entertainment applications, specifically ringtones and games, can be pur-
chased for a one-time download rate. Advanced applications, such as GPS
location services and music and TV applications, are now also available
from some providers—usually for an additional monthly or daily fee.
4. Phones and Lock-In Clauses

Free or discounted phones that come with most postpaid plans add
another dimension of complexity to the cellular product. Different car-
riers offer different phones with varying discounts. The carrier’s choice
between an outright discount and a rebate adds another twist. The flip-
side of the free or discounted phone is the lock-in clause that ties the
consumer to the specific carrier. Lock-in clauses vary in duration and in
the magnitude of the ETF. The common lock-in period is two years, but
one- and three-year periods are also offered. The termination fees vary
between $175 and $200 for standard phones and often reach $350 for
“advanced devices,” such as smartphones. The recent move to graduated
ETFs introduced additional variation as different carriers adopted differ-
ent formulas to govern the gradual reduction in ETFs over the life of the
contract.
5. Prepaid Plans
Not only is it difficult to choose among the many different postpaid plans,
the consumer must make a preliminary choice between postpaid and pre-
paid plans.The prepaid plan is another option offered by the cellular service
market that features a substantially different contractual design than post-
paid plans.
51. For an example of pay-as-you-go pricing—see Verizon Wireless, “Cell Phones—Smart-

phones: Cell Phone Service, Accessories—Verizon Wireless,” above note 39 (as of September
29, 2011,Verizon charged $0.20 per text message and $0.25 per multimedia message).Verizon
also offers fixed-quantity monthly packages (as of September 29, 2011,Verizon charged $5.00
per month for 250 text or multimedia messages and $10.00 for 500 messages). For an exam-
ple of unlimited-quantity monthly packages—see AT&T, “Cell Phones, Cell Phone Plans,
and Wireless Accessories—from AT&T,” above note 39 (AT&T charged $20.00 per month
for unlimited messaging as of September 29, 2011.) Unlimited messaging and even data are
covered by the monthly fee component of the basic three-part tariff in some premium plans.
Id.
c e l l phone s 211
Prepaid offerings fall into two categories: the monthly prepaid category,
in which customers pay a monthly fee for a fixed number of minutes, and
the pay-as-you-go category, in which customers buy credit to pay for min-
utes on a minute-by-minute basis.
The monthly prepaid category more closely resembles the postpaid call-
ing plans. The three main differences are that, unlike postpaid plans, with
prepaid plans:
• The fixed monthly fee is paid in advance.
• There is no commitment.The subscriber can leave the carrier at any time
without incurring an ETF.
• The allotted number of minutes cannot be exceeded, not even for a high
overage charge.
In addition, per-minute prices (the monthly charge divided by the allotted
number of minutes) are generally higher in prepaid plans, perhaps to make
up for the loss of revenue from precluding overage minutes and charges. For
example, for a $25 monthly charge, AT&T’s prepaid GoPhone plan offers
250 minutes ($0.10 per minute), as compared to the 450 minutes for $39.99
offered under AT&T’s postpaid plan ($.09 per minute). Prepaid plans also
offer fewer additional features. For example, night and weekend minutes are
not always unlimited, and roaming charges are levied.52
The second category of prepaid plans offers pay-as-you-go service. Con-
sumers purchase calling cards that hold varying numbers of minutes. For
example, Verizon’s pay-as-you-go service offers prepaid cards with a mini-
mum purchase of $15. These card values translate into calling minutes at a
$0.25 per minute rate. Pay-as-you-go calling cards come with expiration
dates. At Verizon, cards with values of $15 to $29.99 expire in 30 days, cards
with values of $30 to $74.99 expire in 90 days, cards with values of $75 to
$99.99 expire in 180 days, and cards with values of at least $100 expire in 365
days. Verizon’s pay-as-you-go consumers can also pay a fixed fee of $2 to
use the phone for an unlimited number of minutes in a particular day, or use
the phone for a day at a rate of $0.10 per minute. Like the monthly prepaid
plans, pay-as-you-go services typically offer higher per-minute prices and
fewer additional features, as compared to the postpaid plans. Prepaid and
pay-as-you-go plans are not available for smartphone users with any of the
four national carriers.
52. None of the four major operators charges for roaming in its postpaid pricing plans.
III. Explaining the Cellular Service Contract

What explains the contractual design features described in Part II? Why does
the common cellular service contract look the way it does? In this Part, we’ll
explore possible rational-choice, efficiency-based theories for each of the
three design features: three-part tariffs, lock-in clauses, and complexity. But
as we’ll see, these theories provide only a partial account of the contractual
outcomes in the cellular service market. We’ll fill the explanatory gap by
developing a behavioral-economics theory that explains contractual design
in the cellular service market as a market response to consumer mistakes.53
1. Rational-Choice Theories and Their Limits
The leading rational-choice explanation for three-part tariffs views these
pricing schemes as a mechanism for price discrimination or market screening
of rational consumers with different ex ante demand characteristics. For
expositional purposes, let’s focus on two dimensions of demand heterogen-
eity: average (or mean) monthly minutes of use and variance of minutes used.
Suppose that consumers vary only on the first dimension, average
monthly minutes used. Here, the rational model cannot explain three-part
tariffs. To discriminate between heavy users with high average usage and
light users with low average usage, carriers would use a menu of two-part,
not three-part, tariffs. A two-part tariff includes a fixed monthly fee and a
constant per-minute charge. Carriers can discriminate between heavy
users and light users by offering an “H” tariff with a higher monthly fee and
a lower per-minute charge and an “L” tariff with a lower monthly fee and a
higher per-minute charge.The heavy users care more about the per-minute
charge, and will thus prefer the H tariff.The light users care more about the
monthly fee, and will thus prefer the L tariff.
While two-part tariffs provide a mechanism for discriminating between
consumers based on their mean usage, three-part tariffs can provide a mech-
53. The analysis of the three-part tariff, in Section A, relies heavily on Grubb, above note 1. The
behavioral explanation described in Section A is also closely related to the one developed in
K. Eliaz and R. Spiegler, “Contracting with Diversely Naive Agents”, Rev. Econ. Stud., 73
(2006), 689; K. Eliaz and R. Spiegler, “Consumer Optimism and Price Discrimination”,
Theor. Econ., 3 (2008), 459.
c e l l phone s 213
anism for discriminating between consumers based on variance of use.

Assume that there are two types of consumers: one type with highly varia-
ble—or High-Variation (HV)—demand, and another type with more pre-
dictable, Low-Variation (LV) demand.54 Specifically, suppose that the HV
type’s demand will either be H + h or H – h with equal probability, while
the LV type’s demand will be either L + l or L – l with equal probability,
where h > l and H + h > L + l > L – l > H – h. A monopolistic carrier can
discriminate between the HV types and the LV types using a menu of
three-part tariffs. This is because HV types are more concerned than LV
types about: (i) using a very large number of minutes, which makes them
more inclined to choose a tariff with a larger allocation of minutes to reduce
the risk of paying substantial overage fees; and (ii) using only a very small
number of minutes, since then, for any given fixed monthly fee, they end up
paying a higher per-minute price.
The monopolist can exploit this difference by designing a three-part tariff
for LV types with steep overages above L + l, LV’s highest possible demand, and
offering HV types a tariff with at least H + h free minutes for a higher monthly
price. LV types like the LV tariff more than the HV types since: (i) they are
indifferent to the possibility of overages as they never consume minutes above
L + l, while the HV types will end up paying overages half of the time if they
choose the LV tariff, and (ii) the zero marginal price on allotted minutes within
their plan is worth relatively more to LV types, since, unlike the HV types, they
always consume at least L – l of those minutes, whereas HV types still end up
paying a high average price per minute in the event that their demand is low.
While a three-part tariff pricing structure can facilitate price discrimination,
the assumptions required for this rational-choice explanation are often unreal-
istic. In the price-discrimination model, the HV type chooses a plan with a
high number of allotted minutes and the LV type chooses a plan with a low
number of allotted minutes. Moreover, their highly variable use levels imply
that HVs are more likely than LVs to use a very low number of minutes. Using
the data described in subsection 2 below, Figure 4.1 shows the cumulative dis-
tribution functions of usage for consumers choosing different three-part tariff
plans: Plan 1 with 200 included minutes, Plan 2 with 300 included minutes,
Plan 3 with 400 included minutes, and Plan 4 with 500 included minutes.
54. Formally, the cumulative distribution function (“c.d.f.”) describing the priors over the
demand parameter of the predictable type must cross that of the variable type once from
below. Grubb, above note 1.
1.2
0.8
Plan 1
0.6
Plan 2
Plan 3
0.4
Plan 4
0.2
0
0
40
80
120
160
200
240
280
320
360
400
440
480
520
560
600
640
680
720
760
800
840
880
920
960
1000
1040
1080
1120
1160
1200
1240
1280
1320
1360
1400
1440
1480
Minutes
Figure 4.1. Cumulative distribution functions of cell phone usage
Figure 4.1 confirms that the cumulative distribution function corres-

ponding to a plan with a higher number of allocated minutes first-order
stochastically dominates the cumulative distribution function correspond-
ing to a plan with a lower number of allocated minutes.55 In other words,
consumers who choose plans with a higher number of allotted minutes are
less likely to end up using a very low number of minutes.These findings are
inconsistent with the price-discrimination theory.56
An alternative rational-choice explanation is based on risk aversion. In the-
ory, the use patterns revealed in the data (see subsection 2 below) are consistent
with the behavior of perfectly rational but risk-averse subscribers. Such sub-
scribers would choose plans with more allotted minutes than they expect to use
to reduce the risk of paying substantial overage fees. As a result, most of these
subscribers will end up using much less than their allotted minutes.This expla-
nation fails for three reasons. First, given the sums of money involved, the
observed plan choices are not consistent with risk aversion under the rational-
choice Expected Utility Theory.57 Second, as demonstrated below, subscribers often
55. The data aren’t completely consistent with a strict FOSD ranking, however, since Plan 2’s
cumulative distribution function crosses Plan 3’s towards the bottom of the graph. But it is
likely that the Plan 2 data are less reliable.
56. Grubb’s analysis of a different dataset yields the same conclusion. Grubb, above note 1.
57. See Matthew Rabin, “Note, Risk Aversion and Expected-Utility Theory: A Calibration The-
orem”, Econometrica, 68 (2000), 1281. However, they may be consistent with certain behavioral
accounts of risk aversion. See id., at 1282 n. 3.
c e l l phone s 215
choose the wrong plan. And, importantly, these mistakes in plan choice are not
only ex post mistakes as the risk-aversion explanation would imply; many of
them are ex ante mistakes. Finally, while risk aversion may explain the observed
usage patterns, it cannot explain the emergence of the three-part tariff as the
equilibrium pricing structure. With rational, risk-averse subscribers, we should
expect to see two-part tariffs.
2. A Behavioral-Economics Theory
a. Theory
Basic voice services are commonly priced using three-part tariffs.To choose
the optimal three-part tariff from the menu of tariffs offered by the carriers,
consumers must accurately anticipate their future cell phone usage. But
many consumers, when asked to choose a calling plan, are not armed with
accurate estimates of how they will use their cell phones. According to a
pricing manager at a top U.S. cell phone service provider, “people abso-
lutely think they know how much they will use and it’s pretty surprising
how wrong they are.”58 The three-part tariff responds to consumers’ misper-
ceptions about their future use.
Consumers both overestimate and underestimate their use levels. A car-
rier who is aware that consumers suffer from such misperceptions can make
its service plan appear more attractive to consumers than it really is by using
a three-part tariff that charges a low (zero) per-minute price for minutes up
to the plan limit and a high per-minute price thereafter. Consumers who
overestimate their usage overestimate the value of the low prices because
they overestimate the probability that they will use most of these free min-
utes. Conversely, consumers who underestimate their usage pay insufficient
attention to the high overage fees because they underestimate the probabil-
ity of exceeding the plan limit. For a monopolistic carrier, the three-part
tariff creates opportunities for increased profits, while carriers operating in
a competitive market will adopt the three-part tariff because it maximizes
perceived consumer surplus.59
These ideas can be illustrated using a simple numeric example. Assume
that several carriers are operating in a highly competitive market. All carriers
58. Grubb, above note 1.

59. Three-part tariffs can arise also when consumers are overconfident about their ability to pred-
ict their future use. Namely, when the same consumers exhibit a tendency to both over- and
underestimate future use. See Grubb, above note 1.
face the same cost structure: a $10 per-consumer fixed cost and a $0.10 per-
minute variable cost. Consumers have the following preferences: they value
each minute of airtime at $0.40 per minute up to a certain saturation point,
s, while minutes beyond the saturation point are worth zero to the consumer.
There are two types of consumers: heavy users and light users. Fifty percent
are heavy users with a saturation point of 300 minutes, and 50 percent are
light users with a saturation point of 100 minutes. If consumers are rational
and accurately predict their saturation points, then the carriers will set a two-
part tariff with a fixed monthly fee of $10 and a constant, per-minute mar-
ginal price of $0.10. Heavy users will pay 10 + 300 · 0.1 = 40, light users will
pay 10 + 100 · 0.1 = 20, and the carriers will just cover their costs, as expected
in a perfectly competitive market. Under this two-part tariff, heavy users
enjoy a surplus of 300 · (0.4–0.1) –10 = 80 and light users enjoy a surplus of
100 · (0.4–0.1) –10 = 20.60
Let’s introduce consumer misperceptions into the mix. We assume that
light users overestimate their saturation point, mistakenly perceiving a satu-
ration point of 200 minutes instead of the actual 100 minutes. And heavy
users underestimate their saturation point, mistakenly perceiving a satura-
tion point of 200 minutes instead of the actual 300 minutes. With such
misperceptions, a three-part tariff becomes more appealing than the two-
part tariff.
Consider the following three-part tariff: a fixed $10 monthly fee, 200
allotted minutes (at a marginal price of zero), and an overage charge of
$0.40 per minute beyond the 200-minute allocation. The 200-minute allo-
cation tracks the common perceived saturation point, the $0.40 overage is
the maximal marginal price that would not deter usage beyond the plan
limit, and the $10 fixed fee is calculated to exactly cover the carrier’s expected
⎛1 1 ⎞ 1
costs: 10 + ⎜ 2 ⋅ 100 + 2 ⋅ 300⎟ ⋅ 0.1 − 2 ⋅ (300 − 200 )⋅ 0.4 = 10 .61 Under this tar-
⎝ ⎠
60. Price calculations add the fixed monthly fee to the number of minutes multiplied by the per-
minute price. Surplus calculations take the number of minutes multiplied by the difference
between the per-minute benefit and the per-minute price and subtract the fixed monthly
fee.
61. The carrier’s costs include a fixed cost of $10 and an expected variable cost of $0.1 per minute
multiplied by the expected number of minutes—100 minutes for light users (50 percent of
users) and 300 minutes for heavy users (50 percent of users). The total cost is $30. The carrier
gets $20 from overage charges that the heavy users pay on their last 100 minutes. The remain-
ing $10 is collected as a fixed monthly fee.
c e l l phone s 217
iff, heavy users will pay 10+(300–200). 0.4=50.They will enjoy a surplus of
300 0 .04–(300-200). 0.4-10=70, less than the surplus of 80 under the two-
part tariff. But since they underestimate their use, they misperceive the sur-
plus. The perceived surplus under the three-part tariff is 200 0 .04–10=70,
greater than the perceived surplus of 200. (0.4–01)–10=50 under the two-
part tariff. Light users will pay $10 under the three-part tariff. They will
enjoy a surplus of 100 0 .04–10=30, more than the surplus of 20 under the
two-part tariff. More importantly, the perceived surplus under the
three-part tariff is 200 0 .04–10=70, greater than the perceived surplus of
200. (0.4–0.1)–10=50 under the two-part tariff.
The three-part tariff extracts payments in the form of overage fees
that are invisible to consumers,62 while reducing or eliminating pay-
ments that are visible to consumers—specifically, fixed fees and charges
for minutes within the plan limit. Notice that the heavy users, who
underestimate their usage levels and end up paying overage fees, are
subsidizing the light users. But since the heavy users do not anticipate
paying the overage fees, a competitor cannot lure them away ex ante by,
for example, offering a different tariff with lower overage fees. The
three-part tariff maximizes the perceived consumer surplus for both
types of consumers, and thus will be selected as the equilibrium tariff in
a competitive market.
b. Data
A unique dataset of subscriber-level monthly billing and usage information
for 3,730 subscribers at a single wireless provider was used to test the mis-
perception theory.These data provide information on which calling plan (of
four) subscribers chose, as well as monthly consumption of peak minutes for
the period of September 2001 to May 2003. Each of the four calling plans
offered a standard three-part tariff with a fixed allocation of peak minutes
and steep overages for additional peak minutes consumed, as described in
Table 4.1 below.63
62. In a more general model, overage charges would be underestimated, but not completely
invisible.
63. The database was provided by the Center for Customer Relationship Management at Duke
University. The description of the data in the text is based on the description provided by the
Center. See The Center for Customer Relationship Management, Telecom Dataset, available
at <http://www.fuqua-europe.duke.edu/centers/ccrm/index.html#data>; see also Raghuram
Table 4.1. Menu of Three-Part Tariffs

Plan 1 Plan 2 Plan 3 Plan 4
Market share (%) 22.15 2.00 73.28 2.57
Monthly fixed charge ($) 30 35 40 50
Number of included minutes 200 300 400 500
Overage rate ($) 0.40 0.40 0.40 0.40
The data reveal substantial variance in usage. Summary statistics are pro-
vided in Tables 4.2a–e below. For plans 1, 2, 3, and 4, Tables 4.2a–d present
the overall mean and standard deviation of minutes used. To gain an initial
sense of usage underestimation versus overestimation of usage, average fig-
ures for under-usage (unused minutes per month) and over-usage (minutes
over the plan allocation) are also presented. This information is aggregated
across all plans in Table 4.2e.
In aggregate, subscribers exceeded their minute allowance 11.4 percent
of the time by an average of 34.3 percent. In the 88.4 percent of the time
when the allowance was not exceeded, subscribers used, on average, only
45.4 percent of their minute allowance. Notice that “underages” and over-
ages do not, in and of themselves, imply overestimation and underestima-
tion of use. A perfectly rational consumer with variable use will experience
both underages and overages.
Let’s now look at consumers who arguably chose the wrong plan, and
the cost of the mistake to those consumers. A plan choice is a mistake
when, given the consumers’ usage, the selected plan is different from
another available plan that would have cost less. The unit of analysis is the
consumer’s tenure with a plan, and only the 3,456 consumers who stayed
with a plan for at least ten months are considered. Given the variance in
usage from month to month, identifying mistakes over shorter time hori-
zons is less reliable. For each of the 3,456 consumers, the total cost of
Iyengar, Asim Ansari, and Sunil Gupta, “A Model of Consumer Learning of Consumer Serv-
ice Quality and Usage”, J. Mktg. Res., 44 (2007), 529, 535–7. It is not entirely clear from the
data that all four plans were offered at all dates in all markets. We acknowledge this limitation
of the data and qualify our results accordingly. Our empirical strategy builds on Grubb, above
note 1, who tested a related behavioral explanation, the overconfidence theory, using a differ-
ent dataset.
c e l l phone s 219
Table 4.2a. Summary Statistics—Plan 1

Share Usage/Allowance
Mean Std. Dev.
Under allowance 0.819 0.45 0.294
Over allowance 0.179 1.46 0.624
All consumers 1 0.633 0.538
Table 4.2b. Summary Statistics—Plan 2

Share Usage /Allowance
Mean Std. Dev.
Table 4.2c. Summary Statistics—Plan 3

Mean Std. Dev.
Table 4.2d. Summary Statistics—Plan 4

Mean Std. Dev.
Table 4.2e. Summary Statistics—Aggregate

All Plans
Mean Std. Dev.
wireless service under the consumer’s chosen plan was compared to the
total amount that this consumer would have paid had each of the other
three plans been chosen. The magnitude of the mistakes is measured by
the difference (in both percentages and dollars) between the consumer’s
actual wireless costs and the lowest possible cost—the cost that the con-
sumer would have paid if the consumer could have predicted usage with
certainty.64
The results are summarized in Tables 4.3a and 4.3b. In these tables, each
row represents the group of subscribers who chose a certain plan. (Note
that there is no row for Plan 2, since no Plan 2 subscriber remained with the
plan for more than 10 months.) This group is then divided into four sub-
groups according to the plan that these subscribers should have chosen. For
instance, the cell located at the intersection of the Plan 3 row and the Plan
1 column represents the subgroup of subscribers who chose Plan 3 but
should have chosen Plan 1. Table 4.3a presents the size, in percentage terms,
of these subgroups. Table 4.3b presents the magnitude of the mistakes or
cost-savings, both in percentage terms and in annual dollar terms, for each
subgroup.65
The results for one group of subscribers, those who chose Plan 3, are
presented in Figure 4.2. This group of subscribers is noteworthy for its sig-
nificant numbers of both under-estimators, who should have chosen Plan 4,
and over-estimators, who should have chosen either Plan 2 or Plan 1. Figure
4.2 displays the share of Plan 3 consumers who should have chosen each of
the four plans (the dark gray bars). For those who should not have chosen
Table 4.3a. The likelihood of mistakes

Optimal Plan
Chosen plan Plan 1 74.09% 21.79% 1.49% 2.49%
Plan 3 27.20% 35.61% 21.19% 16%
Plan 4 9.00% 10.66% 8.00% 73.33%
64. This analysis assumes risk neutrality. The sums involved are small enough that this assumption
seems reasonable.
65. The magnitude of the mistake (or cost saving) is calculated as follows (this calculation is per-
formed for each cell in Table 4.3b): We calculate the actual total price paid by a consumer
under her chosen plan over the period that she stayed with the plan. We then calculate the
hypothetical total price that this consumer would have paid had she chosen the best plan
given her actual usage.The difference between these actual and hypothetical total prices is the
c e l l phone s 221
Table 4.3b. The magnitude of mistakes

Optimal Plan
Chosen Plan 1 0% 9.56% 26.97% 28.22%
plan $0 $54.16 $203.58 $341.71
Plan 3 21.09% 6.55% 0% 11.34%
$101.58 $32.59 $0 $102.98
Plan 4 36.71% 12.38% 7.00% 0%
$220.27 $75.31 $39.90 $0
Plan 3, Figure 4.2 shows the amount of money they would have saved, both
in percentage terms (the light gray bars) and in dollar figures.
These figures underestimate the number and cost of mistakes, especially for
plans with a lower allocation of minutes. For example, for subscribers who
chose Plan 1, the data only reveal mistakes arising from underestimation of use
(selection of Plan 1) when the subscriber should have chosen Plan 2, Plan 3,
or Plan 4. But it is likely that many Plan 1 subscribers who overestimated their
use could have done better by choosing a prepaid plan that is not included in
the dataset. A conservative estimate of the number and magnitude of the cost
of such overestimation can be generated by adding a hypothetical prepaid
plan with a high per-minute charge of $0.40 (equal to the overage charges in
our data). An estimated 24.4 percent of Plan 1 subscribers would have saved
$149 annually on average had they chosen the prepaid plan.66
The mistakes in plan choice that we have discussed are ex post mistakes.
Since consumers face ex ante uncertainty about their future use, even con-
sumers who forecast their use ex ante in a rational manner will make pre-
dictions that turn out to be incorrect ex post.To assess the extent of ex ante
mistakes in our data and again focusing on subscribers who stayed with the
same plan for at least 10 months, each subscriber’s tenure with the plan was
divided into two: the first-half and the second-half periods.
magnitude of the mistake and, when reporting dollar values, we transform this number into
annual figures (by dividing the difference by the actual number of months under the plan and
then multiplying by 12 months). The mistakes thus calculated are then averaged out across all
consumers in the specific Table 4.3b cell, e.g., across all consumers who chose Plan 3 but
should have chosen Plan 1.
66. These conclusions are tentative, since prepaid plans may differ from postpaid plans on other
dimensions. In particular, while the service quality offered by prepaid plans is improving, in
the period when the data were collected there was still a non-negligible difference in quality
between prepaid and postpaid plans.
40.00%
35.00%
30.00% $101.58
25.00%
20.00%
15.00% $102.97
10.00% $32.59
5.00%
0
0.00%
plan 1 plan 2 plan 3 plan 4
sbscr.% savings%
Figure 4.2. Plan 3 subscribers—likelihood and magnitude of mistakes
Consider a subscriber who, given her usage in the first-half period, should
have chosen Plan 1 but chose Plan 3 instead.This subscriber, who could have
easily switched to Plan 1 at the end of the first-half period (subscribers were
not locked in to a plan), stayed with Plan 3 in the second-half period even
though Plan 1 remained the optimal plan based on the second-half period
usage. Since the same plan was optimal in both half periods, her usage must
have remained roughly the same during the two half periods, which means
that her first-half period usage provided her with a basis to update her pre-
dictions about her usage in the second-half period. Since subscribers were
free to switch plans and their switching costs were likely to be low, we can
surmise that the subscriber’s decision to choose Plan 3 was probably an ex
ante mistake—at least for the second-half period. The data reveal that a sub-
stantial percentage of subscribers who chose the wrong plan in the first-half
period also chose the wrong plan in the second-half period.67
67. This analysis is performed in Oren Bar-Gill and Rebecca Stone, “Pricing Misperceptions:
Explaining Pricing Structure in the Cell Phone Service Market” forthcoming in J. Empirical
Legal Stud.,Vol. 9. Evidence of “bill shock”—a sudden increase in the monthly bill that is not
caused by a change in service plan—also suggests that many consumers are making ex ante
mistakes. A 2010 FCC survey study found that one in six mobile users have experienced “bill
shock,” and that for a sizeable proportion of users the magnitude of the “shock” was substan-
tial—more than a third reported an increase of at least $50, and 23 percent reported an increase
of $100 or more. See FCC, News Release: “FCC Survey Confirms Consumers Experience
Mobile Bill Shock and Confusion about Early Termination Fees,” May 26, 2010, available at
<http://www.fcc.gov/document/fcc-survey-confirms-consumers-experience-mobile-bill
-shock-and-confusion-about-early-termin>; last visited November 17, 2011.
c e l l phone s 223
To sum up: Many consumers fail to accurately anticipate their use

patterns, and the three-part tariff design is a market response to such
misperceptions.
Consistent with this story, providers do not seem to be troubled by con-
sumers’ use-pattern mistakes. In fact, until recently they actively foster these
mistakes by requiring, as a condition for network access, that handset manu-
facturers disable the call-timer feature that would make it easier for con-
sumers to monitor their usage.68 However, consumers are becoming more
aware of their use-pattern mistakes and more frustrated with carriers who
take advantage of them.As elaborated in Part V below, the market is respond-
ing to the demand generated by these more sophisticated consumers.
B. Lock-In Clauses
Lock-in clauses can arise in a rational-choice framework. When the seller
incurs substantial per-consumer fixed costs and the liquidity-constrained
consumer cannot afford to pay an upfront fee equal to these fixed costs,
the optimal solution may be a lock-in contract. In the cell phone market,
fixed costs are high but, more importantly, they are endogenous. Carriers
invest up to $400 in acquiring each new customer.69 Many of these cus-
tomer-acquisition costs, however, are attributed to the free or subsidized
phones that carriers offer.70 This raises several questions. Why do carriers
offer free phones and lock-in contracts? Why not charge customers the
full price of the phone and avoid lock-in? Many cell phone consumers
can afford to purchase the phone up-front. Moreover, it is unlikely that
the carrier is the most efficient source of credit available to all of those
consumers who are in fact liquidity-constrained.Thus, the rational-choice
68. Wu, above note 20, at 9. For an example of the carrier-imposed difficulty customers face in
determining their unused plan-minute allowances—see Sherrie Nachman, “Cranky Consumer:
How to Check Up on Your Cell Phone Minutes,” Wall Street Journal, June 18, 2002, at D2.
69. Lauren Tara Lacapra, “Breaking Free of a Cellular Contract—New Web Sites Help Customers
Swap or Resell Phone Service; Avoiding $175 Termination Fee,” Wall Street Journal, November
30, 2006, at D1 (“It costs a cell phone company approximately $350 to $400 to acquire a new
customer, according to Phil Doriot, a partner in the consulting firm CFI Group, who has
studied company performance and customer satisfaction for major cellular service provid-
ers.”); Jane Spencer, “What Part of ‘Cancel’ Don’t You Understand?—Regulators Crack
Down on Internet Providers, Phone Companies That Make It Hard to Quit,” Wall Street
Journal, November 12, 2003, at D1 (noting that customer acquisition costs are approximately
“$339 per new customer, according to Yankee Group, a technology research firm”).
70. Jared Sandberg, “A Piece of the Business,” Wall Street Journal, September 11, 1997, at R22.
model can explain the presence of these design features in only a subset of
contracts.71
An alternative argument views lock-in clauses as instrumental in stabiliz-
ing demand and helping providers match capacity to demand (especially in
peak hours), thus reducing costs and benefiting consumers. While lock-in
clauses may reduce churn and thus variation in demand, there are still sig-
nificant variations in the use patterns of the locked-in consumers, as shown
above. More importantly, it is not clear that providers need lock-in clauses
to match capacity to demand. Providers have good information about their
customers’ use patterns, including how long they will stay with a specific
provider. A related argument is that ETF-enforced lock-in generates a more
predictable stream of revenues, which is necessary for carriers to recoup
their large capital investments.72 Again, while lock-in reduces uncertainty
for the carriers, these carriers could generate reasonably accurate revenue
estimates without it.Though reduced risk is desirable, the presence of man-
ageable risk should not prevent investment.
The lock-in clauses that are common in postpaid plans and the termin-
ation fees that enforce them can be explained as a market response to the
imperfect rationality of consumers. Consumers often underestimate the
likelihood that switching carriers will be beneficial down the road, because
the service provided by the current carrier is not as good as promised,
monthly charges are higher than expected, or another carrier is offering a
better deal. Since they underestimate the likelihood of eventually wanting
to switch carriers, consumers underestimate the long-term cost of the
lock-in clause. When consumers underestimate the likelihood that they
will want to switch providers before their contract expires, they will be
71. The fact that termination fees were initially structured such that consumers paid the same fee
regardless of when they terminated the contract raises doubts about the argument that ETFs
were necessary to cover the cost of the free or subsidized phones, either by inducing consum-
ers to stay on and pay the monthly subscription fees or by replacing the subscription fees of
consumers who leave.
72. See Thomas J.Tauke, Executive Vice President,Verizon, “Testimony at FCC Early Termination
Hearing” 2 (June, 12, 2008) (hereinafter Verizon Testimony), available at <http://www.fcc
.gov/realaudio/presentations/2008/061208/tauke.pdf>; see also CTIA, “Early Termination
Fees Equal Lower Consumer Rates”, CITA.org, April 2006, <http://files.ctia.org/pdf/
PositionPaper_CTIA_ETF_04_06.pdf>.
c e l l phone s 225
relatively insensitive to the ETF. Increasing the size of the ETF thus
becomes an appealing pricing strategy for carriers. Moreover, the ETF-
enforced lock-in facilitates the common practice of bundling phones and
service. Termination fees guarantee a long-term revenue stream, as sub-
scribers must either refrain from switching carriers and pay for service for
the duration of their contracts or switch and pay the termination fee.73 This
guaranteed revenue enables carriers to offer free or subsidized phones to
attract consumers.
But the story is more complicated. To subsidize the cost of phones, car-
riers must charge an above-cost price for service. This pricing strategy is
attractive only if the price of service is underestimated. As we have seen in
Part III.A, such underestimation is common. Consumers underestimate the
price that they will pay in overage fees when they underestimate usage.
When they overestimate usage, they underestimate the per-minute price
that they will pay under the plan. Of course, a single month’s worth of
underestimated service prices cannot cover the large phone subsidies. Con-
sequently, lock-in is crucial. Lock-in ensures that the carrier will benefit
from (typically) two years’ worth of above-cost and underestimated service
charges or, if lock-in fails, from the underestimated termination fee. These
compounded above-cost service charges can then pay for the free or subsi-
dized phones. Lock-in and bundling also play into consumers’ myopia, fur-
ther compounding the problem. The immediate cost of the phone looms
larger in the decision calculus than the costs of the service contract, which
are spread over time.
Carriers are quite explicit about their strategy of offering free or subsi-
dized phones and recouping their costs through long-term contracts with
ETFs. According to the vice president of marketing for Cingular Wireless
(now AT&T), the penalties are the price that consumers must pay for the
inexpensive or free phones customers get when they sign up for service:
“We subsidize the handset; in exchange we want a commitment from the
73. The ETF effectively deters switching. See Lacapra, above note 69 (according to a July 2005
survey by the U.S. PIRG Education Fund, “[r]oughly 47 percent of cell customers would
switch or consider switching cellphone companies if early-termination fees were abolished,”
but “because of the fee, only 3 percent of customers go ahead with terminating the
contract”).
customer.”74 Similarly, at the FCC hearing on ETFs, an Executive Vice Pres-

ident of Verizon argued:
Term contracts allow the consumer to take advantage of bundled services at
competitive prices and the latest devices they choose in exchange for a com-
mitment to keep the service for usually one or two years. In return, service
providers have some measure of assurance over a fixed period of time that
they may recover their investment, including equipment subsidies, costs of
acquiring and retaining customers, and anticipated revenue for providing
wireless services.75
The pricing of the iPhone is a good example of this strategy in action. In

June 2008, Apple made a big splash when it announced that the new
iPhone model would sell for $200 less than its predecessor ($199 versus
$399). However, at the same time, Apple and its partner AT&T raised the
iPhone’s minimum monthly service subscription from $60 to $70, adding
$240 to the total cost of the two-year contract.76 AT&T and Apple exec-
utives were very clear about the short-term versus long-term trade-off.
They were willing to lose money on the front end, but only because they
were counting on making even more money off the back end, due to the
two-year lock-in contract. Not surprisingly, when the same iPhone was
later offered in unbundled form, without a two-year service plan, it was
priced at $599, which is $400 above the subsidized price (with a service
plan).77
The practice of offering free or subsidized phones with lock-in con-
tracts provides strong evidence of consumer bias. Carriers seem to under-
stand that consumers are attracted by the short-term benefit (the free
phone) even when this benefit is completely offset or even outweighed by
increased long-term costs. While bundling of phones and service is still the
norm in the U.S. cellular service market, this practice seems to be in decline.
Consumers are more aware of ETFs and carriers are reducing ETFs in
74. Caroline E. Mayer, “Griping about Cellular Bills; Differences from ‘Regular’ Phones Take
New Users by Surprise,” Washington Post, February 28, 2001, at G17; see also Fawn Johnson,
“FCC Head Seeks Rules on Cell-Termination Fees,” Wall Street Journal, June 13, 2008, at B7
(“Wireless carriers argue that the termination fees are used to subsidize the cost of cellphones
to customers. People who sign up for one- or two-year contracts receive discounts on phones
and their monthly wireless rates.”); CTIA, above note 72, at 1 (arguing that prohibiting carri-
ers from charging ETFs will cause prices for wireless services to increase).
75. See Verizon Testimony, above note 72, at 1.
76. See Paul Wagenseil, “That ‘Cheaper’ iPhone Will Cost You More,” FoxNews.com, June 11, 2008,
<http://www.foxnews.com/story/0,2933,365347,00.html>.
77. AT&T Plans to Offer No-Contract iPhone, above note 47.
c e l l phone s 227
response—changes that can be attributed, at least in part, to the ETF litiga-

tion. With lower ETFs and thus weaker lock-in, phone subsidies become
more difficult to sustain. The drive towards open access also threatens the
future of the bundling strategy. After initially resisting open access, carriers
are beginning to realize the benefits of shifting development and customer
service costs to handset manufacturers. Finally, it is interesting to note that
the practice of bundling phones and service has always been less common
outside the U.S.78
C. Complexity
The rational-choice explanation for complexity is straightforward. Con-
sumers have heterogeneous preferences. Different consumers want different
kinds of cellular services, so the complexity and multidimensionality of the
cellular service offerings cater to the heterogeneous preferences of cell
phone users. This surely explains some of the observed complexity in the
cell phone market. But it doesn’t fully explain the staggering level of com-
plexity exhibited by the long menus of cell phone contracts. Even for the
rational consumer, acquiring information on the range of complex products
is costly. Comparing different plans with different multidimensional features
is costly, even for this rational consumer. At some point, these costs exceed
the benefits of finding the perfect plan.When complexity deters comparison-
shopping, the benefits of the variety and multidimensionality are left unre-
alized. The rational-choice account must balance the costs and benefits of
complexity. It seems that in the cell phone market the level of complexity
has reached a point beyond what we should expect if it was simply a response
to rational consumer demand.79
78. See Ante, above note 20.

79. A market for “comparison shopping services” is emerging, with vendors such as BillShrink
and Validas offering to find the best product/plan for any consumer who is willing to pay a
fee. See below note 102 and accompanying text. The availability of comparison-shopping
services reduces the cost of comparison-shopping and increases the optimal level of complex-
ity in a rational-choice model. However, it seems that most cell phone users do not avail
themselves of the services offered by BillShrink and Validas. The emergence of a market for
“comparison shopping services” suggests that complexity makes it difficult for consumers to
comparison-shop by themselves. But since the majority of consumers do not seek help from
professional comparison-shoppers and thus do not benefit from the high level of complexity,
the rational choice explanation for complexity is less convincing.
The complexity and multidimensionality of the cell phone contract can be
explained as a market response to the imperfect rationality of consumers.
Consider four basic plans offered by the four major carriers:
(1) AT&T’s $39.99 plan with 450 minutes, $0.45 per minute overage,
unlimited night (9:00 p.m. to 6:00 a.m.) and weekend minutes (which
are, in fact, limited to 5,000 minutes), unlimited calling to AT&T cus-
tomers, and rollover minutes.
(2) Verizon’s $39.99 plan with 450 minutes, $0.45 overage, unlimited
night (9:01 p.m. to 5:59 a.m.) and weekend minutes, unlimited calling
to Verizon customers, and unlimited calling to five “Friends & Family”
numbers.
(3) Sprint’s $39.99 plan with 450 minutes, $0.45 overage, unlimited nights
(7:00 p.m. to 7:00 a.m.) and weekends.
(4) T-Mobile’s $39.99 plan with 500 minutes, $0.45 overage, unlimited calls
to customers on the T-Mobile network, and unlimited nights (9:00 p.m.
to 6:59 a.m.) and weekends.
To choose among these products, the consumer must answer a series of dif-
ficult questions. How important to the consumer is unlimited calling within
the network? If unlimited calling within the network is important, on which
network are most of the consumer’s friends located? How valuable is unlim-
ited calling during weekends? How valuable is unlimited calling at night?
What are the cost implications of unlimited calling at night when “night” is
between 7:00 p.m. and 7:00 a.m. as compared to a shorter “night” between
9:00 p.m. and 6:00 a.m.? How valuable is the rollover feature?
There is considerable complexity even when the comparison is between
Plans 1, 2, and 3, which offer consumers the same monthly charge, number
of allotted minutes, and overage charge. But, of course, the different dimen-
sions of the three-part tariff also change from one carrier to the next and
from one plan to the next in a single carrier’s menu of offerings. Consumers
must choose the combination of monthly charge, allotted minutes, and
overages they prefer. This choice requires accurate estimates of the distribu-
tion of their future usage.
A perfectly informed and perfectly rational consumer could navigate this
maze and find the optimal plan. But the amount of information required to
do so is substantial, since it includes information about both available plans
and the consumer’s own use patterns. The cost of collecting and processing
c e l l phone s 229
all this information may well outweigh the corresponding benefit. Thus,
even a rational consumer will generally be imperfectly informed. For the
imperfectly rational consumer, this imperfect information will also lead to
bias and to systematic underestimation of the total cost of cellular service.
Complexity allows providers to hide the true cost of the contract. Imper-
fectly rational consumers cannot effectively aggregate the costs associated
with the different options and prices in a single cell phone contract. Inevi-
tably, consumers will focus on a subset of salient features and prices and
ignore or underestimate the importance of the remaining, non-salient fea-
tures and prices. In response, carriers will increase prices or reduce the qual-
ity of the non-salient features, which in turn will generate or free up
resources for intensified competition on the salient features. Competition
forces providers to make the salient features attractive and salient prices low.
This can be achieved by adding revenue-generating, non-salient features
and prices. The result is an endogenously derived high level of complexity
and multidimensionality.
Complexity as a response to imperfect rationality is a dynamic process.
Consumer learning implies that a feature or a price that was not salient
last month may become salient next month. ETFs provide such an exam-
ple. When one price dimension becomes salient, competition focuses on
this dimension and carriers shift to a new, less salient price dimension.
According to some accounts, carriers facing increased competition on
fixed monthly fees and allocations of included minutes are now relying
more heavily on revenues from charges for data services.80 The proposed
account of complexity not only allows for consumer learning, but also
uses consumer learning to explain the increasing level of complexity of
the cellular service contract: When consumers learn the importance of a
previously non-salient price dimension, carriers have a strong incentive to
create a new price dimension that will be, at least initially, non-salient.
IV. Welfare Implications

We have seen how the design of cell phone contracts can be explained as
a response to the imperfect rationality of consumers. In this Part, we’ll
assess the extent to which the mistakes that consumers make—and
80. Andrea Petersen and Nicole Harris, “Hard Cell: Chaos, Confusion and Perks Bedevil Wireless
Users,” Wall Street Journal, April 17, 2002, at A1.
providers’ responses to these mistakes—harm consumers and generate wel-

fare costs.
As we have seen, misperceptions of use levels lead many consumers to
choose the wrong plan; more specifically, the wrong three-part tariff.
The average consumer in our data made a mistake that cost 8 percent of the
total wireless bill, or $47.68 annually. Extrapolating from our data
onto the entire U.S. population of cell phone users, which is approximately
280 million, we obtain a $13.35 billion annual reduction in consumer
surplus.
While the $13.35 billion figure is substantial, the average per-consumer
harm, $47.68, seems small. But these averages hide potentially important
distributional implications.The $13.35 billion is not evenly divided among
the 280 million U.S. subscribers. In our data, 35 percent of subscribers
chose the right plan. Even among subscribers who chose the wrong plan,
the magnitude of the mistake—that is, the extra payment as compared to
the right plan—varies substantially. In our data, 34 percent of consumers
made mistakes that cost them at least 10 percent of their total wireless bill,
or $113 annually, and 17 percent of consumers made mistakes that cost
them at least 20 percent of their total wireless bill, or $146 annually. (10
percent of consumers made mistakes that cost them at least 25 percent of
their total wireless bill, or $60 annually; this implies that the really large
mistakes, in percentage terms, had smaller stakes in dollar terms.)
It should be emphasized that a reduction in the consumer surplus is not
a welfare cost in and of itself. Yet the identified consumer mistakes do gen-
erate welfare costs for two reasons. First, consumer mistakes imply ineffi-
cient allocation, since they cause consumers to buy the wrong products.
Second, social welfare is reduced by regressive redistribution. Such redistri-
bution occurs when carriers profit from consumer mistakes (assuming that
carriers’ shareholders tend to be richer than their customers). But regressive
redistribution may occur even if these excess profits are reduced through
competition. The distribution of mistakes implies that revenues from con-
sumers who make mistakes keep prices low for consumers who do not
make mistakes. If the former tend to be poorer than the latter, then wealth
is redistributed in the wrong direction.
c e l l phone s 231
B. Lock-In Clauses
Lock-in prevents efficient switching and thus hurts consumers. A 2005 sur-
vey by the U.S. PIRG Education Fund found that 47 percent of subscribers
would like to switch plans, but only 3 percent do so—the rest are deterred
by the ETF.81 While more recent changes in the structure and magnitude of
ETFs likely resulted in increased switching, current ETFs are still substantial
and still deter switching.
Switching is efficient when a different carrier or plan provides a better
fit for the consumer. In light of the rapid advances in handset technology,
a two-year lock-in is a relatively long period of time. Beyond these effi-
ciency costs, consumers lose from lock-in when it prevents them from
accepting a better deal offered by a competing carrier. Lock-in can also
slow down the beneficial effects of consumer learning. Consumers gradu-
ally learn to avoid misperception and form more accurate estimates of their
future use. If lock-in prevents these consumers from switching to a plan
that better fits their actual use patterns, it prolongs the welfare costs identi-
fied in Part IV.A. Similarly, consumers will gradually learn the implications
of their complex cell phone contract. For example, they may learn that
they do not use their phone very often between 6 a.m, and 7 a.m., and thus
conclude that they are not benefiting from the longer definition of “night”
in Sprint’s unlimited night calling. If lock-in prevents these consumers
from switching to a different carrier, it prolongs the welfare costs of
complexity.
In addition to these direct costs, lock-in may inhibit competition, add-
ing a potentially large indirect welfare cost. We have already mentioned
that lock-in may prevent a more efficient carrier from attracting consum-
ers who are locked into a contract with a less efficient carrier. Since lock-
in makes large-scale entry into the market more difficult, incumbents may
have a greater incentive to seek monopolization through predation or
merger than in markets where easy entry limits incumbents’ market
power.82
81. Lacapra, above note 69.

82. Joseph Farrell and Paul Klemperer, “Coordination and Lock-In: Competition with Switching
Costs and Network Effects” in Armstong, M. and Porter, R. (eds.), Handbook of Industrial Organ-
ization, Vol. 3, pp. 1967, 2005 (North-Holland, 2007).
C. Complexity
The high level of complexity of cellular service contracts can reduce
welfare in two ways. First, consumers will tend to make more mistakes
when the choices are complex. Second, complexity inhibits competition
by raising the cost of comparison-shopping (which discourages com-
parison-shopping). This is true for the perfectly rational consumer;
imperfect rationality only exacerbates the problem. Without the disci-
pline that comparison-shopping enforces, cell phone service providers
can behave like quasi-monopolists, raising prices and reducing consumer
surplus.
D. Countervailing Benefits?
Three-part tariffs, lock-in clauses, and complexity harm consumers and
increase carriers’ profits. Competition among carriers, even if imperfect,
forces carriers to give back to consumers some of these profits. Carriers will
respond to competition by lowering prices that are salient to consumers.
Handset subsidies are the primary way in which benefits flow back to
consumers.
However, these countervailing benefits do not eliminate the identified
welfare costs. Even if all excess profits are returned to consumers, there will
still be an efficiency cost. Consumer mistakes and the contractual design
features that respond to these mistakes lead consumers to misperceive the
relative costs and benefits of different products. As a result, consumers often
choose the wrong products and use these products less optimally than they
otherwise might.
Moreover, even if all excess profits are returned to consumers as a group,
there is no reason to believe that the benefit received by a consumer will
precisely offset the harm to that same consumer. In fact, it is likely that con-
sumers who are more prone to mistakes will be cross-subsidizing consumers
who are less prone to mistakes.The resulting redistribution can reduce social
welfare.
Finally, one important effect of lock-in and complexity is to reduce the
level of competition in the cellular services market. Reduced competition
means that less of the excess profits will find their way back into the hands
of consumers.
c e l l phone s 233
V. Market Solutions
Consumers make mistakes and carriers respond to these mistakes. However,
consumers also learn from their mistakes, and carriers respond to demand
generated by the growing number of increasingly sophisticated consumers.
Indeed, in a competitive market carriers may have an incentive to correct
consumer mistakes—at least when these mistakes prevent consumers from
fully appreciating the benefits of the correcting carrier’s product.
We’ll begin by describing a number of products and contracts that, argu-
ably, respond to demand by more sophisticated consumers.Then, in Section
B, we’ll examine whether these market solutions in fact solve the behavioral
market failures identified in this chapter.
A. Catering to Sophisticated Consumers

The cellular service market boasts a large set of products and contracts that
arguably cater to more sophisticated consumers.
1. Unlimited Calling Plans

In February 2008,Verizon broke with industry pricing norms by offering a
$99 unlimited calling plan. Soon after, AT&T followed suit. T-Mobile went
even further by including unlimited text messaging with unlimited voice in
its unlimited plan. Sprint then unveiled a $99 plan that featured “unlimited
voice, text messages, email, web surfing, video, and other premium serv-
ices.”83 Unlimited calling plans arguably respond to consumer complaints
about overage fees. Most likely, a sufficiently large subset of disgruntled
consumers, experiencing the sting of large overage charges, generated
demand for plans without overage fees.
The rise of unlimited plans demonstrates both the power and potential
unevenness of consumer learning. We have presented the three-part tariff as
a response to consumer misperception about future use. Of the different
components of the three-part tariff, the overage fee is likely to be the one
that consumers learn to appreciate most quickly. When consumers exceed
83. Roger Cheng, “Business Technology: Virgin Mobile to Join Flat Rate Phones Frenzy,” Wall
Street Journal, June 24, 2008, at B4.
the plan limit, they receive direct and painful feedback (an overage fee) that
helps them learn. But as argued earlier, the underestimation of use that
triggers overage charges is just half of the problem. The other half—overes-
timation of use—is more difficult to learn. For a consumer using 50 percent
of the allotted minutes, implying a much higher per-minute rate than ini-
tially expected, there is no direct feedback because the consumer still pays
the same monthly fixed fee. It’s doubtful that many consumers divide this
fee by the number of minutes actually used to derive the real per-minute
price. The result of this uneven learning is unlimited plans, rather than the
optimal two-part tariff pricing scheme.
The currently available unlimited plans are attractive only to a relatively
small fraction of heavy users.With their high monthly fees, the unlimited plans
are less attractive than the standard three-part tariff plans for most users.84
Therefore, the unlimited plans are, at best, a limited market solution targeted at
a small segment of cell phone users.These heavy users may learn more quickly
and demand products that cater to their needs.A more general market solution
to consumer learning about underestimation and overage costs, such as a two-
part tariff, is still absent, as is a market solution to the overestimation problem.
Bundling voice, messaging, and data services in a single “unlimited” plan
with a single monthly fixed fee may be a response to learning of a different
kind. Consumers are confused by complex, multidimensional contracts and are
demanding greater “simplicity.” While a single-price, “unlimited everything”
plan is simpler, its simplicity can be exaggerated. In measuring simplicity, it is
not enough to consider the price and other product attributes of only a single
plan. The level of complexity is a result of the interaction between product
attributes and consumer usage patterns across a carrier’s entire menu of plans.
For example, in order to choose between a $99 unlimited plan and a limited
plan with a lower monthly fee (plus possibly separate charges for text messag-
ing and data services), consumers must still form accurate estimates of their
future use and calculate the expected total price of both plans—no easy task.
2. Rollover Minutes
Consumer use varies from month to month. A consumer may talk 350 min-
utes one month and 550 minutes the next.With a standard 450-minute plan,
this consumer will waste 100 minutes in the first month and pay overage
84. See Jeff Blyskal, “Mostly Talk: New Unlimited Cell Plans Won’t Pay for Most,” Consumer-
Reports.org, February 26, 2008, <http://blogs.consumerreports.org/electronics/2008/02/
mostly-talk-new.html>.
c e l l phone s 235
charges for 100 minutes in the second. With AT&T’s 450-minute plan,
which includes the rollover minutes feature, the 100 spare minutes in the
first month are not wasted. Rather they are “rolled over”—added to the
available minutes for the second month.85 This means that in the second
month, the consumer has 550 minutes instead of 450 minutes and thus will
not pay any overage.86 The rollover feature, which predates the unlimited
calling plans described above, can also be seen as a response to consumer
learning about the costs of underestimated use and overage charges. But
unlike unlimited plans that directly respond to underestimation of use, the
rollover feature seems to respond to a different bias—overconfidence about
use levels, which implies underestimation of use in some months and over-
estimation of use in others. By enabling the consumer to smooth uneven
use over time, the rollover feature mitigates the costs of overconfidence.
3. Prepaid Plans
Prepaid, no-contract plans are the natural choice for a sophisticated con-
sumer who wants flexibility and has learned the costs of lock-in. This flex-
ibility, however, comes at a cost. Not only do prepaid, no-contract subscribers
forgo the phone subsidies offered to postpaid, locked-in subscribers, but
they also pay higher per-minute charges (at least as compared to postpaid
subscribers who use all the allotted minutes under their plans). As a result,
even sophisticated consumers would be reluctant to choose a prepaid plan.
In fact, prepaid, no-contract plans, with their lower profitability, were
designed for distinct segments of consumers—specifically younger and
poorer consumers who have low credit scores and do not qualify for a post-
paid plan.87 The numbers confirm this: In 2008, only 16 percent of U.S. cell
phone users had prepaid plans; among households with incomes above
$75,000, only 6 percent of cell phone users had prepaid plans.88
This is starting to change.With the growth of unlimited prepaid offerings
and the reduction in per-minute rates, prepaid plans are now attracting
85. Unused minutes do not roll over forever. They expire after a year.
86. In this example, the rational non-AT&T customer will switch to a 900-minute plan and pay
an additional $20 per month because this charge is smaller than the average overage paid in
the seemingly cheaper plan: $45/2 months = $22.50.
Respect to Commercial Mobile Services, Twelfth Report” 2297–98 ¶¶ 116–18 (2008), availa-
ble at <http://wireless.fcc.gov/index.htm?job=cmrs_reports>.
88. Opinion Research Corporation, “Prepaid Phones in the U.S.: Myths, Lack of Consumer
Knowledge Blocking Wider Use” 4, 10 (2008), <http://www.newmillenniumresearch.org
/archive/120408_prepaid_myths_survey_report.pdf>.
consumers from segments of the market previously controlled by postpaid

plans. Prepaid is becoming a real alternative to postpaid.89
While having a prepaid alternative is valuable, prepaid plans are not a
panacea. While solving the lock-in problem and avoiding underestimation
of lock-in costs, prepaid plans trigger other consumer mistakes. Mispercep-
tions about future use may still lead consumers to choose the wrong monthly
prepaid or pay-as-you-go plan. In fact, expiration dates on minutes pur-
chased under pay-as-you-go plans—important design features of such
plans—may be a response to consumers’ overestimation of use.
4. Graduated ETFs
As described in Part II.B, carriers have been moving from a time-invar-
iant ETF to a time-variant, graduated ETF structure. This shift is a
response to consumers’ increased awareness and sensitivity to ETFs. The
change in the design of ETF provisions is not a pure market solution.
Rather, it is an example of how consumer learning and legal interven-
tion can work in tandem to change business practices. The ETF story
likely began with a small number of consumers who learned to appreci-
ate the cost of ETFs and initiated litigation against the carriers. The
threat of liability probably pushed carriers to adjust their ETF structure.
But the litigation also facilitated greater awareness and sensitivity to
ETFs among consumers. This adjusted demand was something that car-
riers could not ignore.
5. Open Access
The open-access movement in wireless telecommunications is a market-
driven development that could reduce the costs of lock-in and handset-
service bundling. While carriers are still the leading handset retailers,
recent developments are diminishing their power such that it is likely that
handset manufacturers will increasingly sell their products directly to con-
sumers, who can use the phone on any network. Open access is not a
response to consumer learning about biases and the cost of lock-in. Nev-
ertheless, it is an important development that can reduce the costs of
consumer biases.
89. FCC Fifteenth Report, above note 6, at 67; Jenna Wortham, “Cellphones Without Strings,” New
York Times, February 20, 2009, at B1 (describing the growing attraction of prepaid plans and citing
Pali Research, an investment advisory firm, reporting that, in 2008, sales of prepaid plans grew 13
percent in North America, nearly three times faster than traditional postpaid plans).
c e l l phone s 237
B. Market Solutions and Consumer Welfare

The cellular service market seems quite responsive to demand generated by
increasingly sophisticated consumers who learn from their mistakes. From a
policy perspective, the question is to what extent market solutions mitigate
the welfare costs identified in Part IV. As we have seen, the market promptly
responds when consumers quickly learn about the implications of their
mistakes, as they do when underestimated use leads to overage charges. But
we have also seen that the market responds more sluggishly when learning
is slower because the feedback mechanisms are weaker, as is the case with
overestimated use.
While the market solutions described above have the potential to mini-
mize the welfare costs of the identified behavioral market failure, in practice
their impact is more limited.The reason is that many consumers do not take
advantage of these market solutions. For example, unlimited plans with their
high monthly fees are attractive only to a small fraction of heavy users. Pre-
paid plans are chosen by a minority of consumers. If consumers are not
aware of their mistakes, they will not search for products that reduce the
likelihood and consequences of those mistakes.
It is evident, then, that consumers learn and that the market responds to
the demand generated by these more sophisticated consumers. But welfare
costs are incurred during the interim period. To assess the magnitude of
welfare costs, we need to ascertain the speed of consumer learning and of
the market response to changing demand. Moreover, when consumers learn
to overcome one mistake or when one hidden term becomes salient, carri-
ers have an incentive to add a new non-salient term and to trigger a new
kind of mistake. Even if consumers always catch up eventually, this cat-and-
mouse game imposes welfare costs. Wireless operators are among the lead-
ing generators of consumer complaints.90 Market solutions, while important,
are clearly imperfect.
90. See Spencer E. Ante, “The Call for a Wireless Bill of Rights,” Business Week, March 20, 2008, at
80, available at <http://www.businessweek.com/magazine/content/08_13/b4077080431634
.htm?campaign_id=rss_tech> (noting that, according to the Better Business Bureau, for each
of the past three years, the wireless sector has received more complaints than any other indus-
try). In the fourth quarter of 2011, the FCC received 21,076 complaints about wireless tele-
communications. FCC, Quarterly Report on Informal Consumer Inquiries and Complaints
for Fourth Quarter of Calendar Year 2010 August 15, 2011, available at <http://transition.fcc
.gov/Daily_Releases/Daily_Business/2011/db0815/DOC-309057A1.pdf>.
VI. Policy Implications

The identified behavioral market failure imposes substantial welfare costs.
Consumer learning coupled with market forces works to reduce these wel-
fare costs but do not eliminate them. Can legal intervention help?
In this Part, we’ll focus on disclosure mandates which, we believe, can
help. We’ll start with a brief survey of existing rules and regulations. We’ll
then outline several potential reforms.
A. Existing Regulations
Regulation of the consumer–carrier relationship is largely limited to the
information that the provider must disclose to its consumers.91 The FCC
exercised its powers under the Communications Act by promulgating rules
intended to prevent fraudulent behavior by telecommunications providers
and by increasing the transparency of providers’ billing practices. Providers
must clearly identify the name of the service provider associated with each
billed charge and prominently display a toll-free telephone number that cus-
tomers can call to inquire about or dispute any charges.92 Most import-
antly, since 2005, charges must “be accompanied by a brief, clear,
non-misleading, plain-language description of the service or services ren-
dered” that is “sufficiently clear in presentation and specific enough in con-
tent so that customers can accurately assess that the services for which they
are billed correspond to those that they have requested and received, and
that the costs assessed for those services conform to their understanding of
the price charged.”93 The underlying rationale is “to allow consumers to
better understand their telephone bills, compare service offerings, and
thereby promote a more efficient competitive marketplace.”94 Further disclosure
91. Prohibitions against unfair or deceptive advertising should also be mentioned. On one
important dimension, early termination fees, the law has moved beyond the regulation of
information provided by carriers. See Oren Bar-Gill and Rebecca Stone, “Mobile Mispercep-
tions” Harv. J.L. & Tech. 23 (2009) 49.
92. 47 C.F.R. § 64.2401(a)(1), (d) (2008).
93. 47 C.F.R. § 64.2401(b) (2008).
94. In re Truth-in-Billing and Billing Format, Second Report and Order, Declaratory Ruling, and Second
Further Notice of Proposed Rulemaking, FCC CC Docket No. 98–170, 20 F.C.C.R. 6448, 6450
(2005) (hereinafter Truth-in-Billing 2005).
c e l l phone s 239
requirements are imposed at the state level. In particular, state laws regulate
wireless line item charges—discrete charges that are separately identified on
a consumer’s bill.95
There have been calls for more stringent disclosure requirements. For
instance, in 2003, Senator Charles Schumer introduced a bill—The Cell
Phone User Bill of Rights—designed to improve disclosure and make it
easier for consumers to choose among providers and plans. The bill sought
to ensure that marketing materials and contracts clearly spell out the terms
and conditions of service plans by requiring that all wireless contracts and
marketing materials display a box containing standardized information on a
number of key price dimensions, including the monthly fixed charge, per-
minute charges for minutes not included in the plan, and the method for
calculating minutes charged. Information on included weekday and day-
time minutes and nights and weekend minutes, long-distance charges,
roaming charges, incoming call charges, and charges for directory assistance
would also have to be displayed. Termination and start-up fees and trial
periods would have to be outlined as would any taxes and surcharges. In
addition, the bill would authorize the FCC to monitor service quality
industry-wide and make the resulting data publicly available to enable
consumers to make informed choices among providers.96 The bill has not
been enacted into law.
In 2004, the California Public Utility Commission (CPUC) promul-
gated a similar set of rules. These regulations required wireless providers
and other telecommunications operators to (1) ensure that subscribers
receive clear and complete information about rates, terms, and condi-
tions when customers sign up for the service; (2) produce clearly organ-
ized bills that only contain charges that the subscriber has authorized;
and (3) list all federal, state, and local taxes, surcharges, and fees separately.97
The regulations were suspended by the CPUC less than a year after their
95. Id., at 6462.

96. Cell Phone User Bill of Rights, S. 1216, 108th Cong. (2003). A similar bill, the Wireless Con-
sumer Protection and Community Broadband Empowerment Act, was proposed more
recently by Representative Edward Markey. See Press Release, Office of Rep. Edward Markey,
“Markey Holds Hearings on Draft Bill to Address Wireless Customer Protections,” February
27, 2008, <http://markey.house.gov/index.php?option=com_contentandtask=viewandid=3
281andItemid=241>.
97. See Press Release, California Public Utilities Commission, PUC Sets Protection Rules for
Consumers through Telecommunications Bill of Rights, May 27, 2, <http://docs.cpuc.ca
.gov/published/NEWS_RELEASE/36910.htm>; Robert W. Hahn et al., “The Economics of
‘Wireless Net Neutrality’” J. Competition L. and Econ. 3 (2007), 399, 413.
adoption, after the terms of two commissioners who supported the rules
expired.98 The drive for improved disclosure, however, is continuing:
Twenty-two states have introduced some form of a Cell Phone User Bill
of Rights.99
B. Rethinking Disclosure
1. From Product Attributes to Use Patterns
Consumers in the cellular service market learn, often quite effectively, to
appreciate the implications of their biases and mistakes. Carriers respond
with products that reduce resulting costs to consumers. While these market
solutions are imperfect, the market’s responsiveness suggests that the regula-
tion best suited for the cellular service market would facilitate rather than
inhibit market forces. It is, therefore, not surprising that many of the existing
and proposed laws and regulations focus on the provision of information.
We too focus on rules governing information provision; specifically, on dis-
closure regulation.
Our proposals, however, deviate from existing disclosure regulation and
from other proposals for heightened disclosure regulation in an important
way. Current disclosure regulation focuses on the disclosure of product-
attribute information; in other words, information on the different features
and price dimensions of cellular service. Our proposal emphasizes the dis-
closure of use pattern information—information on how the consumer
will use the product.
The proposed Cell Phone User Bill of Rights illustrates the current
exclusive focus on product-attribute information. It requires comprehen-
sive disclosure of fees and charges. However, a truly informed choice cannot
be based on product attributes alone. To fully appreciate the benefits and
costs of a cellular service contract, consumers must combine product-
attribute information with use-pattern information. To assess the costs of
overage fees, it is not enough to know the per-minute charges for minutes
not included in the plan (as proposed in the bill); consumers must also know
the probability that they will exceed the plan limit and by how much. Like-
wise, to assess the benefit of unlimited night and weekend calling, consum-
98. “California Suspends Wireless Bill of Rights,” ConsumerAffairs.com, January 28, 2005, <http://
www.consumeraffairs.com/news04/2005/cpuc_wireless.html>.
99. See Ante, above note 20.
c e l l phone s 241
ers must also know how many “night” and “weekend” minutes they will use
as well as the precise contractual definition of “night” and “weekend.” Use-
pattern information can be as important as product-attribute information.
The disclosure regime should be redesigned to ensure that consumers have
both categories of information.
2. Disclosing Use-Pattern Information

Conventional wisdom assumes that sellers have better information on prod-
uct attributes while buyers have better information about use patterns. If a
buyer has better information about how she will use the product, then it
makes no sense to require sellers to disclose use-pattern information. The
best that sellers can do is to provide general statistical information on prod-
uct use. The buyer, on the other hand, has specific information on how he
or she, not the average consumer, will use the product—or so the conven-
tional account goes.
While the conventional wisdom is correct in many markets, it is not cor-
rect in the cellular service market. Carriers have valuable statistical use-
pattern information that is not available to subscribers. More importantly,
they have individualized use-pattern data, collected over the course of their
relationships with subscribers. As suggested below, disclosing this informa-
tion can empower consumers and facilitate the efficient functioning of the
cellular service market.
a. Average-Use Disclosures
Carriers collect and analyze enormous amounts of use-pattern information.
They know how the average subscriber will use his or her cell phone. This
statistical use information is not limited to averages taken across the entire
subscriber population. Carriers have, and can be required to disclose, aver-
age-use information for subgroups of consumers who are similar (in terms
of demographic characteristics, product choices made, and so forth) to the
consumer receiving the use-pattern disclosure. As the subgroup over which
the averaging takes place becomes smaller, intra-group heterogeneity
decreases, and the value of the average-use information to the individual
consumer increases. However, excessively small subgroups may also be
undesirable. Averaging over large numbers has the benefit of reducing ran-
domness. Reducing the size of the subgroup reduces this benefit. The opti-
mal size of a subgroup is the product of a tradeoff between the benefit of
reducing heterogeneity and the benefit of reducing randomness.
One potentially beneficial average-use disclosure would target the mis-

perception of use levels that underlies three-part tariffs by requiring carriers
to disclose the average overage charges that consumers pay. Carriers could
also be required to disclose the percentage of consumers who use, for exam-
ple, 50 percent or less of their allotted minutes, or the percentage of con-
sumers who would save money if they switched to a lower fixed-fee,
lower-limit plan. Consumers’ underestimation of the cost of lock-in could
be addressed by requiring carriers to provide information about the per-
centage of consumers who stop using their phones before the end of the
lock-in period but continue paying for them. Carriers could also be required
to disclose the percentage of consumers who broke the contract and paid
the exit penalty.100
b. Individual-Use Disclosures
Despite their potential benefits, average-use disclosures suffer from import-
ant shortcomings. Even when averaging across smaller subgroups of consum-
ers, substantial heterogeneity remains. Heterogeneity limits the value of
average-use information to any individual consumer. Moreover, heteroge-
neity allows optimistic consumers to further discount the value of average-
use information. Most people think that their driving skills are above average,
even though most people cannot be better than average (given a symmetri-
cal distribution of ability about the mean). Similarly, optimistic consumers
might all think that they will never exceed the plan limit, even when pro-
vided with information that the average consumer pays $50 a month in
overage fees.
Fortunately, use-pattern disclosure in the cellular service market need
not be limited to average-use information. The long-term relationship
between carriers and consumers allows for the provision of individualized
use-pattern information.101
Individual-use disclosure can reduce consumers’ misperception of use
levels. Carriers already provide consumers with individualized information
on overage charges. Arguably, this disclosure reduced consumers’ underesti-
mation of use and contributed to the demand to eliminate overage fees—a
100. Both of these disclosures are incomplete measures of the cost of lock-in since they do not
capture consumers who continue using their phones only because they are locked in.
101. Of course, consumers have access to the same use-pattern information. But while providers
save the information and analyze it, consumers tend not to notice it and even if they do
notice it, they tend to forget it.
c e l l phone s 243
demand satisfied by unlimited calling plans. A parallel disclosure would help

reduce the costs consumers incur due to overestimation of use. Carriers
should be required to disclose the number of minutes used and the effective
per-minute price, calculated as the monthly fixed fee divided by the number
of minutes used.
Individual-use disclosure can also help consumers evaluate the costs and
benefits of other plan features. Carriers could be required to disclose the
number of night and weekend minutes used and the costs saved by the
unlimited nights and weekends feature. They could also be required to dis-
close the number of minutes used in in-network calling and the associated
savings. Similarly, Verizon, which offers unlimited calls to five numbers,
could be required to disclose the number of minutes used calling these five
numbers, along with the costs saved by this feature.
The proposed individual-use disclosures should be provided on the
monthly bill and in aggregate form on a year-end summary to account for
month-to-month variations. Lawmakers should also revisit another key fea-
ture of the proposed Cell Phone User Bill of Rights. This bill focuses on
disclosures provided at the time of contracting, which makes perfect sense
when carriers are disclosing product-attribute information. Individual-use
information, on the other hand, is not available to carriers when a new
subscriber signs up for service. Continuous disclosures throughout the life
of the contract are equally important.
3. Combining Use-Pattern Information with Product-Attribute Information

We have focused on the disclosure of use-pattern information as opposed
to product-attribute disclosures. But the more appealing proposals argue
for total cost disclosures, which combine both. For example, the disclosure
of effective per-minute prices combines product-attribute information
(the monthly fixed fee) and use-pattern information (the number of min-
utes used).
Taking total cost disclosure one step further, carriers could be required
to disclose a comprehensive TCO figure for their calling plans—the total
amount paid, or to be paid, by a consumer, including overage charges, on a
yearly basis or over the duration of a plan. For new subscribers, this TCO
figure can be based on average-use information. For existing subscribers,
who are considering whether to renew or switch plans or even switch car-
riers, the TCO figure can be based on individual-use information.
TCO information for a single plan, specifically for the subscriber’s current
plan, may be insufficient. To effectively compare different plans, the sub-
scriber needs TCO information on all plans. Carriers could be required to
provide TCO information for their entire menu of plans or, at least, for sev-
eral main offerings. Perhaps a better solution would be to require carriers to
disclose only the plan with the lowest TCO for the prospective subscriber
and for the existing subscriber whose use patterns have changed. For exam-
ple, the monthly bill or yearly summary can include a notice if an alternative
plan would have a lower TCO than the subscriber’s current plan.102
TCO disclosures are simple and thus useful even to the imperfectly
rational consumer. An alternative paradigm focuses on more complex and
comprehensive disclosures for sophisticated intermediaries, and carriers
rather than consumers. Specifically, carriers could be required to provide
comprehensive use-pattern information in electronic form. Consumers
would not use this information themselves. Rather, they would forward it
to intermediaries that provide comparison-shopping services. Such inter-
mediaries already exist. Companies like BillShrink and Validas promise to
find the right plan for each consumer.103 But they currently do this based
on minimal, usually self-reported, use-pattern information, which, as we
have seen, is often inaccurate. If carriers were required to provide compre-
hensive use-pattern information in electronic form, intermediaries such as
BillShrink or Validas could combine this information with the product-
attribute information that they already have and find the carrier and plan
with the lowest TCO for each individual consumer.
Comprehensive use-pattern information disclosed in electronic form
can be helpful even without the intervention of intermediaries. The con-
sumer could provide the information to competing carriers, soliciting indi-
vidualized TCO figures from each. Consumers could then choose the carrier
and plan with the lowest TCO, given their individual use patterns.
102. Utility companies in Germany have voluntarily adopted an even more pro-consumer policy.
At the end of the year they retroactively match each consumer to the service plan under
which the consumer pays the lowest total price given her use over the past year. See Ian
Ayres and Barry Nalebuff, “In Praise of Honest Pricing”, M.I.T. Sloan Mgmt. Rev. 45 (2003),
24, 27. A similar idea is already being applied by cell phone companies in other countries.
See, e.g., Orange.fr, “Forfait Ajustable Pro,” <http://sites.orange.fr/boutique/files/html/pe_
packpro_forfait_ajustable.html> (Orange in France offers to charge the subscriber at the
end of the month according to the plan that best fits the subscriber’s usage during that
month).
103. “What is BillShrink?,” <http://www.billshrink.com/about/> (last visited September 20, 2011);
“About Validas,” <http://www.validas.com/about.aspx> (last visited September 20, 2011).
c e l l phone s 245
To sum up, consumer choice should be guided by information about the

total cost of the product. Conventional wisdom assumes that consumers
have better information about their own use patterns and thus need only
product-attribute disclosures to calculate total cost.We have seen that carri-
ers may well have better use-pattern information, as well as better product-
attribute information. Carriers can more easily combine the two categories
of information into a total cost disclosure. Alternatively, carriers can provide
comprehensive use-pattern information in electronic form, so that inter-
mediaries, or competing sellers, can use it to calculate TCO figures.
4. Real-Time Disclosure
In addition to disclosures made at the time of purchase and after-the-fact
disclosures in the monthly bill and/or in a year-end summary, individual-use
information can and should be provided in real time. The challenge of keep-
ing track of cumulative use has increased with the invention of multiple-limit
plans. For example, plans with different limits for peak and off-peak minutes
(as well as limits for messaging and data services) have increased the chance
that consumers inadvertently exceed their plan limits. To help consumers
avoid this, carriers should be required to notify their subscribers when they
are about to exceed the plan limit. A consumer receiving such notification
may well decide to cut the conversation short, switch to a land line, or post-
pone the conversation until off-peak hours.
In late 2010, the FCC proposed new regulations that mandated such real-
time disclosures. Concerned about what it terms “bill shock”—unexpected
increases in the monthly bill that come from high roaming fees or exceed-
ing a monthly allotment of voice minutes, texts, or data consumption—the
FCC, in its proposed rules, would require carriers to provide usage alerts in
real time.104 The proposed regulations were put on hold when the carriers,
through their trade group CTIA, agreed to voluntarily implement the real-
time disclosures.105
104. See FCC, News Release: “FCC Proposes Rules to Help Mobile Phone Users Avoid ‘Bill
Shock’,” October 14, 2010, available at <http://www.fcc.gov/rulemaking/10-207>; (last vis-
ited: September 21, 2011). Similar rules already exist in Europe. While some U.S. carriers
voluntarily provide certain usage alerts, the FCC found that “[t]he tools in place to eliminate
bill shock vary widely between service providers and type of service, and can be difficult to
find. Most of the alerts that are offered do not cover all services or are not sent before the
overage charges are incurred.” Id.
105. See FCC Chairman Julius Genachowski, Remarks at Bill Shock Event, Brookings Institu-
tion, Washington, DC, October 17, 2011, available at <http://www.fcc.gov/document
/chairman-genachowski-remarks-bill-shock-event>; (last visited): November 17, 2011).
5. Mobile Disclosure
Traditional disclosure mandates require sellers to provide information
printed on a piece of paper. Mobile technology opens the door to a variety
of additional, more innovative disclosure methods. For example, carriers can
provide information via voice messages, text messages, and even multimedia
messages. These modes of disclosure may be more effective than the trad-
itional paper disclosure.
6. Enhanced Disclosure in Action

Some carriers are already providing product-use information.“Usage analy-
sis” functions are beginning to appear on some carriers’ secured websites.
AT&T’s capped data plans offer usage tracking and alert options. And there
are other examples.106 Certain carriers even provide information that com-
bines product-use and product-attribute information. For instance, AT&T
offers Personal Plan Review, which lets you know if your plan fits your
usage or if you should switch to a different AT&T plan. Moreover, there are
smartphone applications that can track usage. Smartphones can even serve
as virtual intermediaries, providing a recommendation on the best available
plan in the market by combining the subscriber’s individual use information
with rate information available online. These developments should be
applauded. They provide further evidence that market forces can work to
the benefit of consumers. Still, as the FCC observed in the “bill shock”
context, these voluntarily provided, enhanced disclosures vary widely
between service providers and types of service and, in many cases, are insuf-
ficient. A regulatory nudge is probably required.107
Conclusion
The cellular service market, boasting annual revenues exceeding $180 bil-
lion, is one of the largest and most important consumer markets in the
United States. While cell phones provide obvious benefits to consumers,
cellular service contracts, designed in response to consumer biases, hurt
106. See Amy Schatz and Sara Silver, “A Plan to Ease the Shock of Cellphone Bills,” Wall Street
Journal, May 12, 2010; David Pogue, “AT&T’s Capped Data Plan Could Save You Money,”
New York Times, June 3, 2010.
107. Indeed, in the “bill shock” context, such a regulatory nudge, or the threat of a regulatory
nudge, spurred the industry into action. See above Sec.VI.B.4.
c e l l phone s 247
consumers and reduce social welfare. Mistakes in plan choice, triggered by

a key contractual design feature—the three-part tariff—cost consumers
over $13 billion annually. Consumer welfare and market efficiency are fur-
ther reduced by the ETF-enforced lock-in feature and by the sheer com-
plexity of the cell phone contract, which also respond to the imperfect
rationality of consumers. Since consumer mistakes often result from con-
sumers’ misperceptions about their own future use patterns, disclosure man-
dates should require carriers to provide consumers with use-pattern
information. This information should be combined with product-attribute
information in TCO or total-annual-cost disclosures and made available in
electronic, database form to facilitate the work of intermediaries.
Conclusion
W e are all consumers. The contracts discussed in this book are part of
our everyday experience. We enjoy the benefits from the products
and services that these contracts make possible. And we bear the costs cre-
ated by these contracts, when they are designed to exploit our cognitive
biases. Legal policy should strive to minimize the costs, while preserving
the benefits. This book has set out to further our understanding of con-
sumer contracts and to provide guidance for optimal regulation of these
contracts.
I conclude by noting three directions for future research and study:
(1) more contracts;
(2) more countries; and
(3) more regulatory strategies.
After developing, in Chapter 1, a general framework for the study of con-
sumer contracts and their regulation, I proceeded to apply this framework
in three consumer markets—credit cards (in Chapter 2), mortgages (in
Chapter 3), and cell phones (in Chapter 4). The case-study chapters served
a dual purpose: First, they presented a detailed analysis of three important
consumer markets, discussing specific legal policy solutions in each market.
Second, they demonstrated how the general theoretical framework (of
Chapter 1) can be fruitfully applied to important real-world problems.
The second purpose is of particular importance for future research. One
book cannot provide a comprehensive analysis of all consumer contracts.
But it can lay out a general theoretical approach and show how the theory
applies to an important subset of contracts. Future research can then focus
on other consumer markets and contracts. Among the leading candidates
for such future research are insurance contracts and contracts for transporta-
tion services.

conc lus i on 249
The domain of analysis can also be expanded geographically. The in-

depth analysis in the case-study chapters focused on the United States.
While some results from the analysis of the U.S. cell phone market, for
example, are relevant to cell phone markets in other countries, other results
are not. As emphasized at the outset, to fully understand the dynamics of
contract design and the potential role of legal intervention, a market-specific
analysis is required. This is true for the different consumer markets in the
U.S. It is similarly true when moving from, say, the U.S. cell phone market
to the cell phone market in the U.K. (or in the EU).The general framework
developed in this book applies wherever sophisticated sellers design con-
tracts for less sophisticated consumers. But a market-specific analysis is
needed to derive operational results from the general framework.
Finally, future research should explore the range of legal policy strategies
that can be used to correct the behavioral market failure—the distortions
created when sellers design their contracts in response to the imperfect
rationality of their customers.This book focused on one regulatory strategy:
disclosure mandates. In many markets, disclosure is probably the right place
to start. But policymakers concerned about market failure should consider
the full range of available regulatory strategies. Consumer education can
supplement disclosure mandates. Other soft-paternalism strategies include
setting optimally designed default rules and safe harbors. And, in certain
cases, stronger interventions, such as banning of abusive contracts and prac-
tices, may be justified.
Consumer contracts are ubiquitous. They produce substantial benefits,
but can also cause substantial harm. A better understanding of consumer
contracts, and how they can be improved, should be of interest to policy-
makers, as well as to us consumers. This book is an attempt to advance our
understanding of consumer contracts.
Bibliography
Agarwal, S., Amromin, G., Ben-David, I., Chomsisengphet, S., and Evanoff, D. “Can
Mandated Financial Counseling Improve Mortgage Decision-Making? Evidence from
a Natural Experiment.” (2009) Fisher Coll. of Bus. Working Paper No. 2008-03-019.
—— —— “Learning to Cope: Voluntary Financial Education Programs and Loan
Performance During a Housing Crisis.” (2009) Charles A. Dice Center Working
Paper No. 2009-23.
Agarwal, S., Driscoll, J., Gabaix, X., and Laibson, D. “Learning in the Credit Card
Market.” February 8, 2008.
—— —— “The Age of Reason: Financial Decisions over the Life-Cycle and Impli-
cations for Regulation.” (2009) Brookings Papers on Economic Activity. Issue 2.
Driscoll, J., and Laibson, D. “Optimal Mortgage Refinancing: A Closed Form
Solution.” (2007) NBER Working Paper No. 13487.
Ainslie, G. “Derivation of ‘Rational’ Economic Behavior from Hyperbolic Dis-
count Curves.” Am. Econ. Rev, 81(1991). 334.
Amromin, G., Chunyan Huang, J., Sialm, C., and Zhong, E. “Complex Mortgages.”
(2010) FRB of Chicago Working Paper No. 2010-17.
Andrews, E.L. “Fed and Regulators Shrugged as the Subprime Crisis Spread:
Analysis Finds Trail of Warnings on Loans.” New York Times, December 18, 2007.
Angeletos, G.-M., et al. “The Hyperbolic Consumption Model: Calibration,
Simulation, and Empirical Evaluation.” J. Econ. Perspect, 15 (2001). 47.
Ante, S.E. “The Call for a Wireless Bill of Rights.” Business Week, March 20, 2008.
—— “Verizon Embraces Google’s Android.” Business Week,. December 3, 2007.
Apgar, W.C., and Duda, M. Homeownership Pres. Found. “Collateral Damage: The
Municipal Impact of Today’s Mortgage Foreclosure Boom.” (2005).
—— and Herbert, C.E., U.S. Dep’t of Hous. and Urban Dev. “Subprime Lending
and Alternative Financial Service Providers: A Literature Review and Empirical
Analysis.” (2006).
Appelbaum, B., and Nakashima, E. “Banking Regulator Played Advocate over
Enforcer: Agency Let Lenders Grow out of Control, Then Fail.” Wash. Post.,
November 23, 2008.
Armstrong, M. “Interactions between Competition and Consumer Policy.” Compe-
tition Policy International. 4 (2008). 97.
AT&T. Cell Phones, Cell Phone Plans, and Wireless Accessories—from AT&T.
—— Plan Terms.
252 b i bl i og raphy
Ausubel, L.M. Adverse Selection in the Credit Card Market. (1999) 20 (unpublished
manuscript).
—— “Credit Card Defaults, Credit Card Profits, and Bankruptcy.” Am. Bankr. L.J.,
71 (1997). 249.
—— “The Credit Card Industry: A History, by Lewis Mandell.” J. Econ. Lit., 30
(1992). 1517.
—— “The Failure of Competition in the Credit Card Market.” Am. Econ. Rev., 81
(1991). 50.
Avery, R.B., Brevoort, K.P., and Canner, G.B. “Opportunities and Issues in Using
HMDA Data.” J. Real Est. Res., 29 (2007). 351.
Ayres, I., and Nalebuff, B. “In Praise of Honest Pricing.” M.I.T. Sloan Mgmt. Rev., 45
(2003). 24.
Bakos, Y., Marotta-Wurgler, F., and Trossen, D.R. “Does Anyone Read the Fine
Print? Testing a Law and Economics Approach to Standard Form Contracts.”
(2009) NYU Law and Economics Research Paper No. 09-40.
Bar-Gill, O. “Bundling and Consumer Misperception.” U. Chi. L. Rev., 73 (2006).
33.
—— and Board, O. “Product Use Information and the Limits of Voluntary Disclo-
sure.” American Law and Economics Review., 14 (2012). 235.
—— and Bubb, R. “Credit Card Pricing: The CARD Act and Beyond.” forthcom-
ing in Cornell L. Rev.
—— and Ferrari, F. “Informing Consumers About Themselves.” Erasmus Law
Review., 3 (2010). 93.
—— and Stone, R. “Pricing Misperceptions: Explaining Pricing Structure in the
Cell Phone Service Market.” forthcoming in J. Empirical Legal Stud., Vol 9.
—— and Warren, E. “Making Credit Safer.” U. Pa. L. Rev., 157 (2008). 1.
Barr, M.S., Mullainathan, S., and Shafir, E. Behaviorally Informed Financial Services
Regulation. New America Foundation, 2008.
—— —— “Behaviorally Informed Home Mortgage Credit Regulation.” in Belsky,
E.S., and Retsinas, N.P. (eds.). Understanding Consumer Credit. Brooking Press,
2009.
Barr, M.S., Dokko, J.K., and Keys, B.J. “Who Gets Lost in the Subprime Mortgage
Fallout: Homeowners in Low-and Moderate-Income Neighborhoods.” (April
2008).
Beales, H., Craswell, R., and Salop, S. “The Efficient Regulation of Consumer
Information.” J. L. & Econ., 24 (1981). 491.
Been, V. Testimony Before Committee on Oversight and Government Reform
Subcommittee on Domestic Policy. “External Effects of Concentrated Mortgage
Foreclosures: Evidence from New York City.” (May 21, 2008).
Belsky, E.S., and Retsinas, N.P. (eds.). Understanding Consumer Credit. Brooking
Press, 2009.
Ben-Shahar, O., and Schneider, C.E. More Than You Wanted To Know: The Failure of
Mandated Disclosure. Princeton University Press, forthcoming.
b i bl i og raphy 253
Benartzi, S., and Thaler, R.H. “Heuristics and Biases in Retirement Savings Behav-
ior.” J. Econ. Persp., 21 (2007) 81.
Bergstresser, D., and Beshears, J. “Who Selected Adjustable-Rate Mortgages:
Evidence from the 1989–2007 Surveys of Consumer Finances.” (2010) HBS
Working Paper 10-083.
Bernanke, B.S. Speech at the Independent Community Bankers of America Annual
Convention, Orlando, Florida: “Reducing Preventable Mortgage Foreclosures.”
(March 4, 2008).
—— Speech at the Women in Housing and Finance and Exchequer Club Joint
Luncheon, Washington, D.C.: “Financial Markets, the Economic Outlook, and
Monetary Policy.” (January 10, 2008).
—— Testimony Before the Committee on the Budget, U.S. House of Representa-
tives: The Economic Outlook. (January 17, 2008).
Bethel, J.E., Ferrell,A., and Hu, G.“Legal and Economic Issues in Litigation Arising from the
2007–2008 Credit Crisis.” (2008) Harvard Law and Econ. Discussion Paper No. 212.
BillShrink.com. Cell Phone Plans, Compare Best Cellular Service Carrier Deals on
BillShrink.
Blumenstein, R. “The Business-Package Plan: AT&T Sees Wireless as the Key to its
Broader Strategy of Bundling Its Services.” Wall Street Journal, September 20, 1999.
Blyskal, J. “Mostly Talk: New Unlimited Cell Plans Won’t Pay for Most.” Consumer-
Reports.org., February 26, 2008.
Boehm, T.P., and Schlottmann, A. “Mortgage Pricing Differentials Across Hispanic,
African-American, and White Households: Evidence from the American Hous-
ing Survey.” Cityscape: J. Pol’y Dev. and Res., No. 2 (2007). 9.
Bond, P., Musto, D.K., and Yilmaz, B. “Predatory Mortgage Lending.” (2008) FRB
of Philadelphia Working Paper No. 08-24. 183.
Bostic, R.W., et al. “State and Local Anti-Predatory Lending Laws: The Effect of
Legal Enforcement Mechanisms.” J. Econ. and Bus., 60 (2008). 47.
Brevoort, K.P., and Cooper, C.R. “Foreclosure’s Wake: The Credit Experiences of
Individuals Following Foreclosure.” (October 12, 2010).
Brito, D.L., and Hartley, P.R. “Consumer Rationality and Credit Cards.” J. Pol.
Econ’y, 103 (1995). 400.
Bubb, R., and Kaufman, A. “Consumer Biases and Firm Ownership.” (2009)
Working Paper.
Bucks, B., and Pence, K. “Do Borrowers Know their Mortgage Terms?” J. Urban
Econ., 64 (2008). 218.
—— et al. “Changes in U.S. Family Finances from 2004 to 2007: Evidence from the
Survey of Consumer Finances.” Fed. Reserve Bull., vol. 95 (February 2009).
Bureau of the Census. Statistical Abstract of the United States: 2011.
Busse, M.R. “Multi-market Contact and Price Coordination in the Cellular Tele-
phone Industry.” J. Econ. and Mgmt. Strategy, 9 (2008–9) 287.
Cagan, C.L., First Am. CoreLogic, Inc. Mortgage Payment Reset: The Issue and the
Impact. (2007).
Calem, P.S., and LaCour-Little, M. “Risk-Based Capital Requirements for Mort-

gage Loans.” (2001). Bd. of Governors of the Fed. Reserve Sys. Fin. and Econ. Discus-
sion Series Paper No. 2001-60.
—— et al. “Switching Costs and Adverse Selection in the Market for Credit Cards:
New Evidence.” J. Banking and Fin., 30 (2006). 1653.
Camerer, C., et al. “Regulation for Conservatives: Behavioral Economics and the
Case for ‘Asymmetric Paternalism’ “ U. Penn. L. Rev., 151 (2003). 1211.
Campbell, J.Y. “Household Finance.” J. Fin,. 61 (2006). 1553.
CardFlash.
—— “BofA Basic.” September 17, 2009.
—— “BofA Changes.” August 18, 2009.
—— “BofA Clarity.” November 30, 2009.
—— “CARD Act Impact.” November 10, 2009.
—— “Chase Pricing.” November 20, 2007.
—— “Citi Offers Credit with Single APR and No Late Fees.” July 26, 2011.
—— “Consumer Reports.” October 20, 2010.
—— “Debit Cards.” August 18, 2008.
—— “Debit Overdraft.” July 12, 2007.
—— “Debit Rewards.” August 28, 2009.
—— “Dec Card Offers.” December 23, 2008.
—— “Fee Factor 08.” January 14, 2009.–3, 95–6.
—— “Fee Income.” January 11, 2010.–4.
—— “Fees and Recession.” December 19, 2008.
—— “Merchant Fees.” January 13, 2009.
—— “Minimum Payments.” December 17, 2008.
—— “Monthly Payment Rates Edge Up Slightly.” March 18, 2009.
—— “Payment Rates Hit a Four-Year Low in July.” August 27, 2008.
—— “Reward Card Analysis.” June 2, 2008.
—— “Reward Card Review.” June 2, 2008.
—— “Straight Talk.” February 23, 2010.
—— “Visa Debit.” May 5, 2009.
Case, K.E., and Shiller, R.J. “Home Buyer Survey Results 1988–2006.” unpublished
paper,Yale University, 2006.
—— —— “Is There a Bubble in the Housing Market?” (2003) Brookings Papers on
Econ. Activity. No. 2.
—— —— “The Behavior of Home Buyers in Boom and Post-Boom Markets.”
New. Eng. Econ. Rev., (November-December 1988) 28.
Center for Am. Progress, et al. Frequency Questionnaire. (2006).
Center for Customer Relationship Management. Telecom Dataset.
Center for Responsible Lending. “A Snapshot of the Subprime Market.”
—— “Mortgage Lending Overview.”
—— “Priceless or Just Expensive? The Use of Penalty Rates in the Credit Card
Industry.” (2008).
—— “Subprime Spillover: Foreclosures Cost Neighbors $202 Billion: 40.6 Million

Homes Lose $5,000 on Average.” (2008).
—— “What’s Draining Your Wallet? The Real Cost of Credit Card Cash Advances.”
(2008).
Chakravorti, S., and Emmons, W.R. “Who Pays for Credit Cards?” (2001) Fed.
Reserve Bank of Chi. Emerging Payments Occasional Paper Series. No. 1.
Chan, S., Gedal, M., Been, V., and Haughwout, A. “The Role of Neighborhood
Characteristics in Mortgage Default Risk: Evidence from New York City.” (2011)
NYU Working Paper.
Chang, R. “Proof that Handset Brands Help Sell Wireless Plans.” RCR Wireless.
October 28, 2008.
Cheng, R. “Business Technology:Virgin Mobile to Join Flat Rate Phones Frenzy.”
Wall Street Journal. June 24, 2008.
Ching, A., and Hayashi, F. “Payment Card Rewards Programs and Consumer Pay-
ment Choice.” (2008).
Civils, W., and Gongloff, M. “Subprime Shakeout: Lenders that Have Closed Shop,
Been Acquired or Stopped Loans.” Wall Street Journal Online.
Collins, J.M. “Education Levels and Mortgage Application Outcomes: Evidence of
Financial Literacy.” (2009) Institute for Research on Poverty Discussion Paper No.
1369-09.
Comptroller of the Currency. “Advisory Letter: Credit Card Practices.” (September
14, 2004).
—— “Guidance on Unfair or Deceptive Acts or Practices.” Advisory Letter No. AL
2002-3 (March 22, 2002).
—— Truth in Lending: Comptroller’s Handbook. (2006).
Congressional Budget Office. The Budget and Economic Outlook: Fiscal Years 2008 to
2018. (2008).
ConsumerAffairs.com. “California Suspends Wireless Bill of Rights.” January 28, 2005.
Consumers Union. Women in the Subprime Market. October 2002.
Courchane, M.J., Surette, B.J., and Zorn, P.M. “Subprime Borrowers: Mortgage
Transitions and Outcomes.” J. Real Est. Fin. and Econ., 29 (2004). 365.
Craswell, R. “Taking Information Seriously: Misrepresentation and Nondisclosure
in Contract Law and Elsewhere.” Va. L. Rev., 92 (2006). 565.
Credit Suisse, Mortgage Liquidity du Jour: Underestimated No More. (2007).
Crockett, R.O. “The Last Monopolist.” Business Week, April 12, 1999.
CTIA. “Early Termination Fees Equal Lower Consumer Rates.” CITA.org. April
2006.
De Long, J.B., et al. “Noise Trader Risk in Financial Markets.” J. Pol. Econ’y, 98
(1990). 703.
—— —— “Positive Feedback Investment Strategies and Destabilizing Rational
Speculation.” J. Fin., 45 (1990). 379.
DellaVigna, S., and Malmendier, U.“Contract Design and Self-Control:Theory and
Evidence.” Quar. J. Econ., 119 (2004). 353.
—— —— “Paying Not to Go to the Gym.” Amer. Econ. Rev., 96 (2006). 694.

Dēmos, “The Plastic Safety Net: The Reality Behind Debt In America.” (2005).
Demyanyk,Y., and Van Hemert, O. “Understanding the Subprime Mortgage Crisis.”
Rev. Fin. Stud., 24 (2011). 1848.
Department of Business Innovation and Skills and Cabinet Office Behavioural
Insights Team. Better Choices: Better Deals., April 13, 2011.
Douglass, E. “The Cutting Edge Special Report: Wireless Communications;
“Pre-paid.” Idea is Catching On in U.S. Market.” L.A.Times, March 15, 1999.
Draut, T., and Silva, J. Borrowing to Make Ends Meet: The Growth of Credit Card Debt
in the ’90s. Dēmos, 2003.
Duhigg, C. “What Does Your Credit-Card Company Know About You?” New York
Times, May 17, 2009.
Durkin, T.A. “Consumers and Credit Disclosures: Credit Cards and Credit Insur-
ance.” Fed. Reserve Bull., (April 2002).
—— “Credit Cards: Use and Consumer Attitudes, 1970-2000.” Fed.Reserve Bull.,
vol. 98 (September 2000).
Edwards, C., and Crockett, R.O. “New Music Phones—Without the i.” Business
Week., April 16, 2007.
Eggert, K. “Limiting Abuse and Opportunism by Mortgage Servicers.” Housing
Pol’y Debate, 15 (2007). 753.
Eliaz, K., and Spiegler, R. “Consumer Optimism and Price Discrimination.” Theo-
retical Economics, 3 (2008). 459.
—— —— “Contracting with Diversely Naïve Agents.” Rev. Econ. Stud., 73 (2006). 689.
Elliehausen, G., Staten, M.E., and Steinbuks, J. “The Effect of Prepayment Penalties
on the Pricing of Subprime Mortgages.” J. Econ. and Bus., 60 (2008). 33.
Ellis, D. “The Effect of Consume Interest Rate Deregulation on Credit Card Vol-
umes, Charge-Offs and the Personal Bankruptcy Rate.” (1998) 98-05 Bank Trends.
Ellison, G. “A Model of Add-On Pricing.” Q.J. Econ., 120 (2005). 585.
—— “Bounded Rationality in Industrial Organization.” in R. Blundell, W. K.
Newey and T. Persson (eds.). Advances in Economics and Econometrics: Theory and
Applications, Ninth World Congress.Vol. II, ch. 5 (2006).
—— and Ellison, S.F. “Search, Obfuscation, and Price Elasticities on the Internet.”
Econometrica, 77 (2009). 427.
Elstrom, P. “Mike Armstrong’s Strong Showing.” Business Week, January 25, 1999.
—— “Wireless With All The Trimmings.” Business Week, November 16, 1998.
Engel, K.C., and McCoy, P.A. “A Tale of Three Markets: The Law and Economics
of Predatory Lending.” Tex. L. Rev., 80 (2002). 1255.
—— —— “Turning a Blind Eye: Wall Street Finance of Predatory Lending.” Ford-
ham L. Rev., 75 (2007). 2039.
Epstein, R.A. “Second-Order Rationality.” in E. J. McCaffery and J. Slemrod (eds).
Behavioral Public Finance, (2006) 355.
Eskridge, W.N., Jr. “One Hundred Years of Ineptitude: The Need for Mortgage
Rules Consonant with the Economic and Psychological Dynamics of the Home
Sale and Loan Transaction.” Va. L. Rev., 70 (1984). 1083.
Evans, D.S., and Schmalensee, R. Paying with Plastic. MIT Press, 1999. 211.
Family Hous. Fund. “Cost Effectiveness of Mortgage Foreclosure Prevention:
Summary of Findings.” (1998).
Farrell, J., and Klemperer, P.“Coordination and Lock-In: Competition with Switch-
ing Costs and Network Effects.” (2007) in Armstrong, M., and Porter, R. (eds.).
Handbook of Industrial Organization.Vol. 3 (2007).
FCC.“Annual Report and Analysis of Competitive Market Conditions With Respect
to Commercial Mobile Services, Fifteenth Report.” (2011) FCC 11-103.
—— “Annual Report and Analysis of Competitive Market Conditions With
Respect to Commercial Mobile Services, Thirteenth Report.” (2009) 24
F.C.C.R. 6185.
Respect to Commercial Mobile Services, Twelfth Report.” (2008) FCC
08-28.
Respect to Commercial Mobile Services, Eleventh Report.” (2006) 21
F.C.C.R. 10947.
Respect to Commercial Mobile Services, Tenth Report.” (2005) 20 F.C.C.R.
15908.
—— News Release: “FCC Proposes Rules to Help Mobile Phone Users Avoid
‘Bill Shock’.” October 14, 2010.
—— “Notice of Inquiry: Consumer Information and Disclosure.” CG Docket No.
09-158, released August 28, 2009.
—— “Quarterly Report on Informal Consumer Inquiries and Complaints for
Fourth Quarter of Calendar Year 2010.”August 15, 2011.
Federal Reserve Bank of Boston. “Consumer Behavior and Payment Choice: 2006
Conference Summary.” (2007) Public Policy Discussion Paper 07-4.
Federal Reserve Board. “Credit Card Lending: Account Management and Loss
Allowance Guidance.” SR Letter 03-1. January 8, 2003.
—— “Credit Cards—Fees.”
—— “Credit Cards—Interest Rates.”
—— “Profitability of Credit Card Operations of Depository Institutions.”
—— “Report to the Congress on Practices of the Consumer Credit Industry in
Soliciting and Extending Credit and their Effects on Consumer Debt and Insol-
vency.” June 2006. 108.
—— “Report to the Congress on the Profitability of Credit Card Operations of
Depository Institutions.” June 2009.
—— “What You Need to Know: New Credit Card Rules Effective February 22.”
—— “What You Need to Know: New Overdraft Rules for Debit and ATM
Cards.”
—— and U.S. Dep’t of Hous. and Urban Dev. Joint Report to the Congress Concerning
Reform to the Truth and Lending Act and the Real Estate Settlement Procedures Act.
(1998).
—— et al. Interest-Only Mortgage Payments and Payment-Option ARMs—Are They for

You? (2006).
Feinberg, R.A. “Credit Cards as Spending Facilitating Stimuli: A Conditioning
Interpretation.” J. Consumer Res., 12 (1986). 384.
Fishbein, A.J., and Woodall, P. “Women Are Prime Targets for Subprime Lending:
Women Are Disproportionately Represented in High-Cost Mortgage Market.”
(2006).
Ford, G.S., Koutsky, T.M., and Spiwak, L.J. “Wireless Net Neutrality: From Carter-
fone to Cable Boxes.” Phoenix Ctr. Pol’y Bull., 17 (April 2007). 2.
Forrester, J.P. “Still Mortgaging the American Dream: Predatory Lending, Preemp-
tion, and Federally Supported Lenders.” U. Cin. L. Rev., 74 (2006). 1303.
Fox, J.T. “Consolidation in the Wireless Phone Industry.” (2005) Net Inst. Working
Paper No. 05-13.
Frank, J.M. “Dodging Reform: As Some Credit Card Abuses Are Outlawed, New
Ones Proliferate.” Center for Responsible Lending, 2009.
Fratantoni, M., et al. Mortgage Bankers Ass’n. The Residential Mortgage Market and
Its Economic Context in 2007. MBA Research Monograph Series, 2007.
Freddie Mac. Automated Underwriting: Making Mortgage Lending Simpler and Fairer for
America’s Families. (1996).
—— “Half of Subprime Loans Categorized as ‘A’ Quality.” Inside B&C Lending.
June 10, 1996.
Frederick, S., et al. “Time Discounting and Time Preference: A Critical
Review.”J. Econ. Lit., 40 (2002). 351.
Frontline. “The Credit Card Game.” November 24, 2009.
Furletti, M. “Credit Card Pricing Developments and Their Disclosure.” (2003) Fed.
Res. Bank of Philadelphia Payment Cards Center Discussion Paper.
Gabaix, X., and Laibson, D.“Shrouded Attributes, Consumer Myopia, and Informa-
tion Suppression in Competitive Markets.” Q.J. Econ., 121 (2006). 505.
Genachowski, J. Remarks at Bill Shock Event, Brookings Institution, Washington,
DC. October 17, 2011.
General Accounting Office, Report to the Chairman and Ranking Minority
Member, Special Committee on Aging, U.S. Senate.“Consumer Protection: Fed-
eral and State Agencies Face Challenges in Combating Predatory Lending.”
Gao-04-280 (2004).
Gerardi, K., et al. “Making Sense of the Subprime Crisis.” Fed. Reserve Bank of
Boston. (2008) Public Policy Discussion Paper No. 09-1.
—— —— “Subprime Outcomes: Risky Mortgages, Homeownership Experiences,
and Foreclosures.” (2007) Fed. Reserve Bank of Boston Working Paper No. 07-15.
Gilder, G. “The Wireless Wars.” Wall Street Journal, April 13, 2007.
Gilo, D., and Porat, A. “The Hidden Roles of Boilerplate and Standard-Form Con-
tracts: Strategic Imposition of Transaction Costs, Segmentation of Consumers,
and Anticompetitive Effects.” Michigan Law Review, 104 (2006). 983.
Gilovich, T., et al. (eds.). Heuristics and Biases: The Psychology of Intuitive Judgment.
Cambridge University Press, 2002.
Glaeser, E.L. “Psychology and the Market.” Amer. Econ. Rev. Papers & Proceedings, 94
(2004) 408.
Government Accountability Office. Credit Cards: Increased Complexity in Rates and
Fees Heightens Need for More Effective Disclosures to Consumers. (2006).
—— Report to the Committee on Financial Services, House of Representatives.
“Real Estate Brokerage: Factors That May Affect Price Competition.” GAO-05-
947 (2005).
Gramlich, E.M., Subprime Mortgages: America’s Latest Boom and Bust. Urban Institute
Press, 2007.
Green, R.K., and Wachter, S.M. “The American Mortgage in Historical and Inter-
national Context.” J. Econ. Persp., 19 (2005). 93.
Grubb, M.D. “Selling to Overconfident Consumers.” Amer. Econ. Rev., 99 (2009).
1770.
Gruber, J., and Koszegi, B. “Is Addiction ‘Rational’? Theory and Evidence.” (2000)
NBER Working Paper No. 7507.
Guiso, L., Sapienza, P., and Zingales, L. “Moral and Social Constraints to Strategic
Default on Mortgages.” (2009) NBER Working Paper No. 15145.
Hahn, R.W., et al. “The Economics of ‘Wireless Net Neutrality’.” J. Competition L.
and Econ., 3 (2007). 399.
Haughwout, A., Mayer, C., and Tracy, J. “Subprime Mortgage Pricing: The Impact
of Race, Ethnicity, and Gender on the Cost of Borrowing.” (2009) FRB of New
York Staff Report No. 368.
Heidhues, P., and Koszegi, B. “Exploiting Naiveté about Self-Control in the Credit
Market.” Amer. Econ. Rev., 100 (2010). 2279.
Helft, M., and Labaton, S. “Google Pushes for Rules to Aid Wireless Plans.” New
York Times, July 21, 2007.
Hernandez-Murillo, R., Ghent, A.C., and Owyang, M. “Race, Redlining, and Sub-
prime Loan Pricing” (2011).
Hirschman, E.C. “Differences in Consumer Purchase Behavior by Credit Card
Payment System.” J. Consumer Res., 6 (1979). 58.
Hoak, A.,”100 Percent More Difficult: First-Time Home Buyers Struggle to Find
Down-Payment Money.” MarketWatch, March 9, 2008.
Immergluck, D., and Smith, G. “The Impact of Single-Family Mortgage Foreclos-
ures on Neighborhood Crime.” Housing Stud., 21 (2006). 851.
Iyengar, R., Ansari, A., and Gupta, S. “A Model of Consumer Learning of Con-
sumer Service Quality and Usage.” J. Mktg. Res., 44 (2007). 529.
Jackson, T.H. The Logic and Limits of Bankruptcy Law. Harvard University Press,
1986.
Jensen, R. “The Digital Provide: Information (Technology). Market Performance,
and Welfare in the South Indian Fisheries Sector.” Q.J. Econ,. 122 (2007). 879.
Johnson, F. “FCC Head Seeks Rules on Cell-Termination Fees.” Wall Street Journal,
June 13, 2008.
Jolls, C., et al. “A Behavioral Approach to Law and Economics.” Stan. L. Rev., 50
(1998). 1471.
Kahneman, D., Slovic, P., Tversky, A., (eds.) Judgment under Uncertainty: Heuristics and
Biases. Cambridge University Press, 1982.
Kjos, A. “Proposed Changes to Regulation Z: Highlighting Behaviors that Affect
Credit Costs.” (2008) FRB of Philadelphia Payment Cards Center Discussion Paper
No. 08-02.
Klein, L. It’s In The Cards: Consumer Credit and the American Experience. Praeger, 1999.
Koopmans, T.C. “Stationary Ordinal Utility and Impatience.” Econometrica, 28
(1960). 287.
Korobkin, R. “Bounded Rationality, Standard Form Contracts, and Unconsciona-
bility.” U. Chi. L. Rev., 70 (2003) 1203.
Kronman, A.T.“Mistake, Disclosure, Information, and the Law of Contracts.” J. Legal
Stud., 7 (1987). 1.
Kroszner, R.S. Speech at the George Washington University School of Business
Financial Services Research Program Policy Forum: “Creating More Effective
Consumer Disclosures” (May 23, 2007).
Lacapra, L.T. “Breaking Free of a Cellular Contract—New Web Sites Help Cus-
tomers Swap or Resell Phone Service; Avoiding $175 Termination Fee.” Wall
Street Journal, November 30, 2006.
Lacko, J.M., and Pappalardo, J.K. Fed.Trade Comm’n.“Improving Consumer Mort-
gage Disclosures: An Empirical Assessment of Current and Prototype Disclosure
Forms.” (2007).
—— —— “The Effect of Mortgage Broker Compensation Disclosures on Con-
sumers and Competition: A Controlled Experiment.” (2004).
LaCour-Little, M. “Economic Factors Affecting Home Mortgage Disclosure Act
Reporting.” J. Real Est. Res., 29 (2007). 479.
—— and Holmes, C. “Prepayment Penalties in Residential Mortgage Contracts: A
Cost-Benefit Analysis.” Housing Pol’y Debate., 19 (2008). 631.
Laibson, D., et al. “Self-Control and Saving for Retirement.” Brookings Papers Econ.
Activity, 1 (1998). 91.
Larson, M.D. “Mortgage Lenders Want a Commitment—and They’re Willing to
Pay You for It.” Bankrate.com, August 26, 1999.
Lax, H., et al. “Subprime Lending: An Investigation of Economic Efficiency.” Hous-
ing Pol’y Debate, 15 (2004). 533.
Lazich, R.S. (ed.). Market Share Reporter: An Annual Compilation of Reported Market
Share Data on Companies, Products, and Services: 2008 (2008).
Lee, J., and Hogarth, J.M. “Consumer Information Search for Home Mortgages:
Who, What, How Much, and What Else?” Fin. Services Rev., 9 (2000). 277.
—— —— “The Price of Money: Consumers’ Understanding of APRs and Con-
tract Interest Rates.” J. Pub. Pol’y and Marketing, 18 (1999). 66.
Leland, J. “Baltimore Finds Subprime Crisis Snags Women.” New York Times, January
15, 2008.
Liebowitz, S.J. “Anatomy of a Train Wreck: Causes of the Mortgage Meltdown.” in
Holcombe, R.G., and Powell, B. (eds.). Housing America: Building Out of a Crisis.
The Independent Institute, 2009.
Loewenstein, G.F., and O’Donoghue, E. “Animal Spirits: Affective and Deliberative

Processes in Economic Behavior.” (2004).
Lusardi, A. “Americans’ Financial Capability.” (2011). NBER Working Paper
17103.
—— “Financial Literacy: An Essential Tool for Informed Consumer Choice?”
(2008) Nat’l Bureau of Econ. Research Working Paper No. 14084.
—— “Household Saving Behavior: The Role of Financial Literacy, Information,
and Financial Education Programs.” (2008) Nat’l Bureau of Econ. Research Working
Paper No. 13824.
—— and Mitchell, O.S. “Planning and Financial Literacy: How Do Women Fare.”
McCoy, P.A. “Rethinking Disclosure in a World of Risk-Based Pricing” Harv. J. on
Legis., 44 (2007). 123.
MacDonald, D.A. “Viewpoint: Card Industry Questions Congress Needs to Ask.”
American Banker, March 23, 2007.
McGeehan, P. “Soaring Interest Compounds Credit Card Pain for Millions.” New
York Times, November 21, 2004.
Macro International. “Design and Testing of Effective Truth in Lending Disclo-
sures.” (2007).
Madison, M.T., Dwyer, J.R., and Bender, S.W. The Law of Real Estate Financing.
Thomson Reuters/West, rev. ed. 2008.
Madrian, B.C., and Shea, D. “The Power of Suggestion: Inertia in 401(k) Participa-
tion and Savings Behavior.” Quart. J. Econ., 116 (2001). 1149.
Mandell, L. The Credit Card Industry: A History. Twayne Publishers, 1990.
Mann, R.J.“Adopting, Using, and Discarding Paper and Electronic Payment Instru-
ments:Variation by Age and Race.” (May 1, 2011).
—— “Bankruptcy Reform and the ‘Sweat Box’ of Credit Card Debt.” U. Ill. L.
Rev., (2007). 375.
—— Charging Ahead: The Growth and Regulation of Payment Card Markets Around the
World. Cambridge University Press, 2006.
—— “ ‘Contracting’ For Credit.” Mich. L. Rev., 104 (2006). 899.
Martin, A. “Credit Card Industry Aims to Profit From Sterling Payers.” New York
Times, May 19, 2009.
Massey, C., and Thaler, R.H. “The Loser’s Curse: Overconfidence vs. Market Effi-
ciency in the National Football League Draft.” (2010).
Massoud, N.Z., Saunders, A., and Scholnick, B. “The Cost of Being Late: The Case
of Credit Card Penalty Fees.” (2006) AFA 2007 Chicago Meetings Paper.
Mayer, C.E. “Griping About Cellular Bills; Differences From ‘Regular’ Phones Take
New Users by Surprise.” Washington Post, February 28, 2001.
Mayer, C.J., and Pence, K.M. “Subprime Mortgages: What, Where, and to Whom?”
Mayer, C.J., Pence, K.M., and Sherlund, S. “The Rise in Mortgage Defaults.” (2008)
Bd. of Governors of the Fed. Reserve Sys. Fin. and Econ. Discussion Series Paper No.
2008-59.
—— and Sinai, T. “Housing and Behavioral Finance” (2007) (unpublished

manuscript).
—— Piskorski, T., and Tchistyi, A. “The Inefficiency of Refinancing: Why Prepay-
ment Penalties Are Good for Risky Borrowers.” (2011) Columbia Business School
Working Paper.
Medoff, J., and Harless, A. The Indebted Society: Anatomy of An Ongoing Disaster. Little
Brown & Co., 1996.
Meier, S., and Sprenger. C. “Present-Biased Preferences and Credit Card Borrow-
ing.” Amer. Econ. J., 2 (2010). 193.
Mester, L.J. “Why Are Credit Card Rates Sticky?” Econ.Th., 4 (1994). 505.
Meyer, D. “Coverage Problems Trigger Headaches for Carriers.” RCR Wireless
News, July 9, 2001.
Mian, A.R., and Sufi, A. “Household Leverage and the Recession of 2007-09.” IMF
Econ. Rev., 58 (2010). 74.
Miles, D.“The U.K. Mortgage Market:Taking a Longer-TermView, Interim Report:
Information, Incentives and Pricing.” (2003).
Miles,W.“Boom-Bust Cycles and the Forecasting Performance of Linear and Non-
Linear Models of House Prices.” J. Real Est. Fin. and Econ., 36 (2008). 249.
Morgenson, G. “Inside the Countrywide Lending Spree.” New York Times, August
26, 2007.
—— “Countrywide Subpoenaed by Illinois.” New York Times, December 13, 2007.
—— “Clicking the Way to Mortgage Savings.” New York Times, December 23,
2007.
—— “Given a Shovel, Digging Deeper Into Debt.” New York Times, July 20, 2008.
—— “S.E.C. Accuses Countrywide’s Ex-Chief of Fraud.” New York Times, June 4,
2009.
Nachman, S. “Cranky Consumer: How to Check Up on Your Cell Phone Min-
utes.” Wall Street Journal, June 18, 2002.
Nalebuff, B., and Ayres, I. Why Not? Harvard Business School Press, 2003.
National Consumer Law Center. Report, Fee-Harvesters: Low-Credit, High-Cost Cards
Bleed Consumers. (2007).
—— Truth in Lending. 27 (2002 Cumulative Supplement).
Nat’l Cmty. Reinvestment Coal. “Homeownership and Wealth Building Impeded:
Continuing Lending Disparities for Minorities and Emerging Obstacles for
Middle-Income and Female Borrowers of All Races.” (2006).
Nuechterlein, J.E., and Weiser, P.J. Digital Crossroads. MIT Press, 2005.
Odlyzko, A.M. “The Many Paradoxes of Broadband.” First Monday. September 1,
2003.
O’Donoghue, E., and Rabin, M. “Doing It Now or Later.” Amer. Econ. Rev., 89
(1999). 103.
—— —— “Procrastination in Preparing for Retirement.” in Aaron, H. (ed.). Behav-
ioral Dimensions of Retirement Economics. Brookings Institution Press, 1999.
Opinion Research Corporation. “Prepaid Phones in the U.S.: Myths, Lack of Con-
sumer Knowledge Blocking Wider Use.” (2008).
Parker, P.M., and Röller, L.H. “Collusive Conduct in Duopolies: Multi-Market

Contact and Cross-Ownership in the Mobile Telephone Industry.” Rand J. Econ.,
28 (1997). 304.
Paulson, H.M., Jr., U.S. Sec’y of the Treasury. “Remarks on Current Housing and
Mortgage Market Developments at the Georgetown University Law Center.”
(October 16, 2007).
Pavlov, A. and Wachter, S. “Sub-prime Lending and Real Estate Prices.” Real Estate
Econ. 39 (2011). 1.
Peek, J. “A Call to ARMs: Adjustable Rate Mortgages in the 1980s.” New Eng. Econ.
Rev. March-April (1990) 53.
Pennington-Cross, A., and Ho, G.“The Termination of Subprime Hybrid and Fixed
Rate Mortgages.” (2006) Fed. Reserve Bank of St. Louis, Research Div. Working
Paper No. 2006-042A.
Petersen, A., and Harris, N. “Hard Cell: Chaos, Confusion and Perks Bedevil Wire-
less Users.” Wall Street Journal, April 17, 2002.
Peterson, C.L. “Preemption, Agency Cost Theory, and Predatory Lending by Bank-
ing Agents: Are Federal Regulators Biting Off More Than They Can Chew?”
Am. U. L. Rev., 56 (2007). 515.
PEW Charitable Trusts. “Safe Credit Card Standards.” (2009).
—— “Safe Credit Cards Project: Curing Credit Card Penalties.” (2009).
Pogue, D. “AT&T’s Capped Data Plan Could Save You Money.” New York Times,
June 3, 2010.
Prelec, D., and Simester, D. “Always Leave Home Without It: A Further Investiga-
tion of the Credit-Card Effect on Willingness to Pay.” Marketing Letters, 12
(2001). 5.
Privacy Rights Clearinghouse/UCAN. “Paper or Plastic? What Have You Got to
Lose.” (2011).
Prudential Ins. Co. of Am. Financial Experience and Behaviors Among Women.
(2006).
Quercia, R.G., Stegman, M.A., and Davis, W.R. “The Impact of Predatory Loan
Terms on Subprime Foreclosures: The Special Case of Prepayment Penalties and
Balloon Payments.” Housing Pol’y Debate, 18 (2007). 311.
Rabin, M. “Risk Aversion and Expected-Utility Theory: A Calibration Theorem.”
Econometrica, 68 (2000). 1281.
Rappaport, T. Wireless Communications. Prentice Hall, 1996.
Reardon, M. “Unlocking the Unlocked Cell Phone Market.” CNET News,. July 2,
2009.
Renuart, E. “An Overview of the Predatory Mortgage Lending Process.” Housing
Pol’y Debate, 15 (2004). 467.
—— and Thompson, D.E. “The Truth, The Whole Truth, and Nothing but the
Truth: Fulfilling the Promise of Truth in Lending.” Yale J. on Reg,. 25 (2008). 181,
188.
Rose, M.J. “Origination Channel, Prepayment Penalties, and Default.” forthcoming
in Real Estate Economics.
Rotemberg, J. Subprime Meltdown: American Housing and Global Financial Turmoil.

Harvard Business School, 2008.
Rubin, E.L. “Legislative Methodology: Some Lessons from the Truth-in-Lending
Act.” Geo. L.J., 80 (1991). 233.
Samuelson, P. “A Note on Measurement of Utility.” Rev. Econ. Stud., 4 (1937). 155.
Sandberg, J. “A Piece of the Business.” Wall Street Journal, September 11, 1997.
Schatz, A., and Silver, S. “A Plan to Ease the Shock of Cellphone Bills.” Wall Street
Journal, May 12, 2010.
Schuh, S.D., and Stavins, J. “Summary of the Workshop on Consumer Behavior and
Payment Choice.” (2008) FRB of Boston Public Policy Discussion Paper No. 08-5.
Schwartz, A. “Who Takes Out Adjustable Rate Mortgages?” (2009) Harvard Univer-
sity Working Paper.
Schwartz, A. and Wilde, L.L.“Imperfect Information in Markets for Contract Terms:
The Examples of Warranties and Security Interests.” Va. L. Rev., 69 (1983). 1387.
—— —— “Product Quality and Imperfect Information.” Rev. Econ. Stud., 52 (1985).
251.
Schwartz, M., Mobile Wireless Communications. Cambridge University Press, 2005.
Semeraro, S. “The Reverse-Robin-Hood-Cross-Subsidy Hypothesis: Do Credit
Card Systems Effectively Tax the Poor and Reward the Rich?” Rutgers L. J., 40
(2009). 419.
Shah Goda, G., Flaherty Manchester, C., and Sojourner, A.J. “What’s My Account
Really Worth? The Effect of Lifetime Income Disclosure on Retirement Sav-
ings.” (2011) RAND Working Paper No.WR-873.
Sharma, A., and Cheng, R. “iPhone Costs Prove a Drag for AT&T.” Wall Street
Journal, October 23, 2008.
—— and Searcey, D. “Verizon to Open Cell Network to Others’ Phones.” Wall
Street Journal, November 28, 2007.
Sherlund, S.M. “The Past, Present, and Future of Subprime Mortgages.” (2008) Bd.
of Governors of the Fed. Reserve Sys. Fin. and Econ. Discussion Series Paper No.
2008-63.
Shiller, R.J. Irrational Exuberance. Princeton University Press, 2000.
—— “Speculative Prices and Popular Models.” J. Econ. Persp., Spring (1990). 58.
—— “Understanding Recent Trends in House Prices and Home Ownership.”
(2007) Yale Univ. Econ. Dep’t Working Paper No. 28.
Shleifer, A., and Summers, L.H. “The Noise Trader Approach to Finance.” J. Econ.
Persp., 4 (1990). 19.
Shroder, M.D. “The Value of the Sunshine Cure: The Efficacy of the Real Estate
Settlement Procedures Act Disclosure Strategy.” Cityscape: J. Pol’y Dev., and Res.
No. 1(2007). 73.
Shui, H., and Ausubel, L.M. “Time Inconsistency in the Credit Card Market.”
(2004) Working Paper.
Sichelman, L.“Community Group Claims CitiFinancial Still Predatory.” Origination
News, January 2002.
Souleles, N.S. “Do Liquidity Constraints and Interest Rates Matter for Consumer
Behavior? Evidence from Credit Card Data.” Quar. J. Econ., 117 (2002). 149.
Spencer, J. “What Part of ‘Cancel’ Don’t You Understand?—Regulators Crack
Down on Internet Providers, Phone Companies That Make It Hard to Quit.”
Wall Street Journal, November 12, 2003.
Spiegler, R. Bounded Rationality and Industrial Organization. Oxford University Press,
2011.
—— “Competition over Agents with Boundedly Rational Expectations.” Theoreti-
cal Econ., 1 (2006). 207.
Sprint. “Cell Phones, Mobile Phones, and Wireless Calling Plans from Sprint.”
SRI International. “The Role of NSF’s Support of Engineering in Enabling Tech-
nological Innovation, Final Report Phase II 94-97.” (1998).
Stallings, W. Wireless Communications and Networking. Prentice Hall, 2002.
Stango, V. “Pricing with Consumer Switching Costs: Evidence from the Credit
Card Market.” J. Indus. Econ., 50 (2002). 475.
—— and Zinman, J. “Fuzzy Math, Disclosure Regulation and Credit Market Out-
comes.” (2007) Tuck Sch. of Bus.Working Paper No. 2008-42.
—— —— “What Do Consumers Really Pay on Their Checking and Credit Card
Accounts? Explicit, Implicit, and Avoidable Costs.” Amer. Econ. Rev., 99 (2009).
424.
Stanovich, K.E. “The Fundamental Computational Biases of Human Cognition:
Heuristics that (Sometimes) Impair Decision Making and Problem Solving.” in
Davidson, J.E., and Steinberg, R.J. (eds.). The Psychology of Problem Solving. Cam-
bridge University Press, 2003.
Stein, E. Coal. for Responsible Lending. “Quantifying the Economic Costs of
Predatory Lending.” (2001).
Sullivan, T.A., Warren, E., and Westbrook, J.L. As We Forgive Our Debtors: Bankruptcy
and Consumer Credit in America. Oxford University Press, 1989.
—— —— The Fragile Middle Class: Americans in Debt. Yale University Press, 2000.
Sunstein, C.R. “Disclosure and Simplification as Regulatory Tools.” Memorandum
for the Heads of Executive Departments and Agencies (2010).
—— “Informing Consumers through Smart Disclosure.” Memorandum for the
Heads of Executive Departments and Agencies (2011).
—— and Thaler, R.H. “Libertarian Paternalism is not an Oxymoron.” U. Chi. L.
Rev., 70 (2003). 1159.
Svenson, O. “Are We All Less Risky and More Skillful than Our Fellow Drivers?”
Acta Psychologica, 47 (1981). 143.
T-Mobile, Financial Release. “T-Mobile USA Reports Second Quarter 2011
Results.” July 28, 2011.
—— “Cell Phones / 4G Cell Phone Plans / Android Tablet PCs / T-Mobile.”
—— “T-Mobile Terms and Conditions”
Tauke, T.J., Executive Vice President,Verizon. Testimony at FCC Early Termination
Hearing ( June, 12, 2008).
Thaler, R.H. “Mental Accounting Matters.” J. Behav. Decision Making., 12 (1999).

183.
—— “Some Empirical Evidence on Dynamic Inconsistency.” Econ. Letters., 8 (1981).
201.
—— The Winner’s Curse: Paradoxes and Anomalies of Economic Life. Princeton Uni-
versity Press, 1992.
—— and Sunstein, C.R. Nudge: Improving Decisions about Health, Wealth and Happi-
ness. Yale University Press, 2008.
Tversky, A., and Kahneman, D. “Rational Choice and the Framing of Decisions.” in
R. M. Hogarth and M. W. Reder (eds). Rational Choice:The Contrast Between Eco-
nomics and Psychology. University of Chicago Press, 1987.
U.S. Dep’t of Hous. and Urban Dev. Report to Congress on the Root Causes of the
Foreclosure Crisis. (2010).
—— “HUD Proposes Mortgage Reform to Help Consumers Better Understand
Their Loan, Shop for Lower Costs.” (March 14, 2008).
—— and U.S. Dep’t of the Treasury. “Curbing Predatory Home Mortgage Lend-
ing.” (2000).
U.S. Dep’t of Justice and Fed. Trade Comm’n. “Competition in the Real Estate
Brokerage Industry: A Report by the Federal Trade Commission and the U.S.
Department of Justice.” (2007).
Vascellaro, J.E. “Air War: A Fight Over What You Can Do on a Cell-phone.” Wall
Street Journal., June 14, 2007.
Verizon Wireless. “Cell Phones—Smartphones: Cell Phone Service, Accesso-
ries— Verizon Wireless.”
—— “Customer Agreement.”
—— “Verizon Wireless—4Selected Financial Results.” ( July 22, 2011).
Wagenseil, P. “That ‘Cheaper’ iPhone Will Cost You More.” FoxNews.com, June 11,
2008.
Waverman, L., Meschi, M., and Fuss, M. “The Impact of Telecoms on Economic
Growth in Developing Countries.” in The Vodafone Policy Paper Series no. 3, Africa:
The Impact of Mobile Phones. (2005).
Weeks, W. “An Analysis and Critique of Retroactive Penalty Interest in the Credit
Card Market.” (2007).
Weinstein, N.D. “Unrealistic Optimism About Future Life Events.” J. Personality &
Soc. Psychol., 39 (1980). 806.
White, M.J. “Bankruptcy Reform and Credit Cards.” J. Econ. Persp., 21 (2007).
175.
Willis, L.E. “Decisionmaking and the Limits of Disclosure: The Problem of Preda-
tory Lending: Price.” Md. L. Rev., 65 (2006). 707.
Woodward, S.E. “Consumer Confusion in the Mortgage Market.” (2003).
—— U.S. Dep’t of Hous. and Urban Dev. A Study of Closing Costs for FHA Mort-
gages. (2008).
Wu, T. “Wireless Net Neutrality: Cellular Carterfone and Consumer Choice in

Mobile Broadband.” (2007). New Am. Found. Wireless Future Program Working
Paper No. 17.
Zinman, J. “Debit or Credit?” (2006) Dartmouth College Economics Department Work-
ing Paper.
Zywicki, T.J. “The Economics of Credit Cards.” Chap. L. Rev., 3 (2003). 79.
—— and Adamson, J.D. “The Law and Economics of Subprime Lending.” U. Colo.
L. Rev., 80 (2009). 1.
Index
Introductory Note
References such as “178–9” indicate (not necessarily continuous) discussion of a topic
across a range of pages. Wherever possible in the case of topics with many references,
these have either been divided into sub-topics or only the most significant discussions
of the topic are listed.
ability-to-repay requirement 147, 184 more inclusive 126–7, 181–2

abusive lending practices 104, 158, 167 purchase 66, 96–7
actual benefits 9–10, 44–5, 48 Ante, S.E. 201, 203–4, 227, 237, 240
actual prices 3, 9–10, 15, 23, 25, 44–5, applications market 204–5
112 appraisal fees 118, 177–8
add-ons 188, 208–10 appraisals 20, 142, 154, 178
adjustable-rate mortgages see ARMs APRs see annual percentage rates
affordability 119, 122, 146–50, 158 arbitration clauses, mandatory 74–5
Agarwal, S. 26, 29, 79–80, 101, 144, arbitration fees 119, 144
160–2, 165 ARMs (adjustable-rate mortgages)
allotted minutes 28, 186–7, 205, 208–9, 117–18, 137–8, 142, 150, 153–4,
213–14, 228, 234–5 163, 169
alternative mortgage products 130, option 138–9, 142, 145, 148, 158–9
147–8, 177 payment-option 118, 123, 147, 158
American Express 58, 63–4, 103 asymmetric paternalism 32
amortization, negative 118, 139, 142, 157 AT&T 63, 185, 194, 197–8, 205–10,
annual fees 14, 18, 47–9, 69, 76, 92, 94–7 225, 246
zero 49, 92 attribute information see
annual percentage rates (APRs) 38–9, 66, product-attribute information
95, 97, 106, 125–7, 174–84 Ausubel, L. 58–9, 63–4, 70–1, 82, 86,
calculation 112, 125–7, 176, 179, 92–3, 113–14
182, 184 average-use disclosure 35, 241–2
definition 127, 181, 184 average-use information 35 , 107,
disclosure 39, 111, 125–6, 174–80, 241–3
182–4
failure 177–80 back-end costs 98, 104
fixing 180–3 balance transfers 70, 73–6, 102, 113
promise of 174–6 fees 52, 66, 95
and subprime balloon payments 169, 172
mortgages 174–83 bankruptcies 51, 58, 60, 62–3, 88–9,
effective 67, 109, 178 99, 165
270 i ndex
Bar-Gill, O. 20, 30–1, 34, 41, 97, imperfect rationality of 52–4, 78–9,
103, 134 82, 112
behavior-based fees 76, 95 misperceptions 55–6
behavioral-economics theory 2, 7–23, 32, optimistic 22, 54, 82, 89
43, 81, 128, 156, 188 cards
cellular service contracts 215–29 charge 57–8
and complexity 79–81, 158–60 credit see credit cards
credit cards 52–3, 56, 66, 78–97, debit 56–7, 101, 103–5
104–5, 114 merchant 57–8
and heterogeneity in cognitive payment 57, 61
ability 160–4 carriers 185–93, 199–206, 208–12, 215–16,
and market correction 164–5 223–34, 236–8, 240–7
subprime mortgages 121, 123, 156–65 major 191, 201, 204, 207–9, 228
behavioral economics theory, common regional 196, 198–9
design features 17–23 switching 224–5
behavioral market failures 2–4, 16–17, 32, CBO see Congressional Budget Office
42–3, 65, 112, 237–8 cell phones 1, 4–5, 27–8, 35, 185–247
benefits applications market 204–5
actual 9–10, 44–5, 48 bundling of phones and service 19–20,
perceived 9–10, 16, 44–5, 47–50 185, 187, 206, 225–7
short-term 1, 20, 53, 82, 175, 226 cellular service see cellular service
Bergstresser, D. 160, 163 handsets see handsets
Bernanke, B.S. 121, 135, 137–8, 169–70 history 195–6
Beshears, J. 160, 163 market 35, 44, 189, 191, 193, 194–9, 227
bias 2–3, 8–9, 16–17, 24–6, 101–3, 158, consolidation 195–6, 199–200
164–5 economic significance 196–8
cognitive 22, 53, 107, 248 open access 236
consumer 6, 8, 54, 101, 226, 236, 246 rise of 194–8
optimism 22, 82, 88, 121 technology 194–5
underestimation 54, 91, 113 cellular service 19–20, 185–6, 190–1,
biased consumers 25, 50, 176 194–5, 201, 206, 227
design of contracts for 8–14 carriers see carriers
bill shock 222, 245–6 contracts 40, 185, 187, 189, 192, 205–11,
borrowers 76, 89, 101, 116–22, 124–8, 130, 229
137–84 add-ons 209–10
imperfect rationality of 121, 124, allotted minutes 28, 186–7, 205,
174–5, 180, 183 208–9, 213–14, 228, 234–5
bounded rationality 2–3, 36 complexity 208–11, 227–9
brokers 130, 132–3, 158–9, 165, 173, 180 behavioral-economics
Bubb, R. 97, 102 theory 228–9
bubbles 152–3 rational choice theories 227
bundling 19–20, 185, 187, 206, 225–7 welfare implications 232
existing regulations 238–40
Cagan, C.L. 127, 138–9, 170, 182–3 explanation 212–29
calling plans see cellular service, contracts family plans 188, 209
card issuers 63–4, 73, 95–6 graduated early termination fees
cardholders 53–4, 57, 67–8, 71–7, 79–82, (ETFs) 192, 201, 207, 210, 236
90–4, 97–9 see also credit cards; lock-in clauses 185, 187, 189, 206–7,
debit cards 210, 223–7, 231–2
i nde x 271
behavioral-economics subscribers 186–7, 189–90, 205,

theory 224–7 217–18, 220–2, 230–1, 244–6
rational choice theories 223–4 Center for Responsible Lending 74,
welfare implications 231 79–80, 97, 116
market solutions 191–2, 233–7 CFPB see Consumer Financial Protection
and consumer welfare 237 Bureau
pay-as-you-go 209, 211, 236 charge cards 57–8
peak minutes 1, 217 charges 18, 30, 49, 58, 69, 234–5,
policy implications 192–3, 238–40
238–47 monthly 186–7, 205, 211, 224, 228
postpaid plans 188, 205–6, 208–11, overage 191, 216–17, 221, 228, 233–5,
221, 224, 235–6 237, 242–3
prepaid plans 188, 190, 198, 210–11, closed-end loans 77, 86, 126
221, 235–7 cognitive biases 22, 53, 107, 248
rational-choice theories 188–9 collective-action problem 30–1
rethinking of disclosure 240–7 comparison-shopping 24, 65, 126, 176–7,
roaming charges 206, 211, 239 180–1, 189–90, 232
rollover minutes 188, 228, 234–5 cost of 3, 132, 190, 227, 232
three-part tariffs 185–9, 191, 205–6, effective 3, 23, 98, 124, 166, 181
208, 212–23, 230, 232–3 competition 16–17, 23–4, 64–5, 93–6,
behavioral-economics 131–3, 199–202, 229–32
theory 215–23 cellular service market 199–202
rational choice theories 212–15 and complexity 190, 232
welfare implications 230 credit card market 64–5, 98–9
two-part tariffs 212, 215–17, 234 distorted 23–5, 98–9, 102, 167
unlimited plans 191, 208–9, 228, enhanced 124, 131, 133, 176, 191
233–5, 237, 243 imperfect 132, 191
weekend minutes 1, 211, 228, 239, inhibition of 23–4
241, 243 limited 23, 98, 132, 166–7
welfare costs 189–91 limits of 16–17
welfare implications 229–32 subprime mortgages market 131–3
countervailing benefits 232 competitive markets 2, 15–16, 19, 31,
disclosure 48–9, 165, 215–17
average use 241–2 complexity 18–21, 23–6, 52–4, 117–22,
combining use-pattern information 158–60, 187–90, 227–9
with product-attribute and behavioral-economics
information 243–5 theory 79–81, 158–60
enhanced disclosure in and cellular service contracts 190,
action 246–7 208–11, 227–9, 232
individual use 242–3 and competition 190, 232
mobile 246 and credit cards 66–8, 75–7, 79–81
real-time 245 and rational choice theories 75–7
rethinking of 240–7 and subprime mortgages 141–5,
use-pattern information 241–3 158–60
market 35, 185, 188, 191, 198–205, 212, comprehensive disclosure 37, 43, 240, 244
240–2 Congress 39, 51, 59, 107–8, 134–5,
competition 199–202 168–9, 195
related markets 202–5 Congressional Budget Office
structure 198 (CBO) 116, 129, 159, 164
272 i ndex
consumer bankruptcies see bankruptcies mortgages 117–23

consumer contracts 1–2, 5–8, 17–18, and salience 92–4
43–4, 204, 248–9 see also inefficient 16, 98–9, 117
contracts loan 86, 117, 121, 125, 135–6, 140, 142
design 2–3, 43 long-term 187, 206–7, 225
consumer credit 51, 57, 60–2, 71, 87, 135 mortgage see mortgages, contracts;
consumer debt 51, 60, 62–3, 87, 108 subprime mortgages, contracts
Consumer Financial Protection Bureau optimal 81, 156, 167–8
(CFPB) 40, 55, 125, 135, 175, subprime mortgages see subprime
178, 183–4 mortgages, contracts
consumer-friendly credit cards 56, 102–3 convenience fees 74, 91, 95
consumer heterogeneity 17, 35, 37, 39, cost deferral 20, 78, 111–12, 121–2, 124–5,
76, 155, 172 146–7, 184–5 see also deferred
consumer markets 5–6, 22, 33–4, 42–4, costs
246, 248–9 cost of credit 73, 106–7, 111, 126, 151,
consumer welfare 16, 190, 237, 247 176, 181
consumers 1–5, 7–30, 34–51, 82–95, costs
100–15, 185–93, 212–49 back-end 98, 104
biased see biased consumers of lock-in 235–6
imperfect rationality of 4, 187–9, 224, long-term 1, 39, 112, 120, 158,
228–9, 247 175–6, 185
imperfectly informed 9, 33, 45 credit 59–62, 64, 71–2, 102–3, 142–3, 171,
learning 26–30, 188, 190–2, 218, 229, 174
231, 233–8 cost of 73, 106–7, 111, 126, 151, 176,
misperceptions 15, 17, 20, 30–1, 45, 181
105–6, 186 total cost of 112, 126, 174, 178, 181
mistakes 32, 38, 165, 189–90, 212, 230, credit cards 1, 4–6, 10–12, 14–15, 26–9,
232–3 33, 47–115
psychology 2–4, 6–8, 15, 17, 26, 32, contracts 14, 36, 65–76, 78, 80–3,
43–4 90–3, 98
rational 2–5, 9–10, 16–21, 23–5, 47–9, behavioral-economics theory 52–3,
187–9, 227–9 56, 78–97, 104–5, 114
contingent fees 144, 182 CARD Act 96–7
contracts 1–2, 4–8, 14–18, 48–50, 80–2, complexity 79–81
184–92, 248–9 deferred costs 81–91, 156–8
cell phones 186–8 salience 91–6
cellular service see cellular service, complexity 66–8
contracts consumer-friendly 56, 102–3
complex 19–20, 23–4, 98, 145, 166, deferred costs 68–74
179, 190 design 52–4
consumer see consumer contracts and distorted competition 98–9
credit card see credit cards, contracts distributional concerns 55, 98,
deferred costs see deferred costs 100–1, 112
design 2–3, 6, 14–15, 28, 50, 117, 124 and hindered competition 98
behavioral-economics theory market solutions 101–5
of 21, 56 policy implications 55–6
for biased consumers 8–14 and rational choice theories 75–8
credit cards 52–4 complexity 75–7
features 2, 17, 28, 75, 97, 124, 136 deferred costs 77–8
i nde x 273
and underestimation of cost of slow and behavioral-economics

repayment 89–90 theory 81–91, 156–8
and underestimation of future credit cards 68–74, 77–8, 81–91
borrowing 87–8 long-term costs 70–4
and underestimation of risk of short-term benefits 68–70
economic hardship 88–9 deferred-cost contracts 21, 38–9, 81–2,
welfare implications 54–5, 97–101 121, 146–7, 156–7, 175 see also
disclosure of product-use cost deferral
information 106–7 and rational choice theories 77–8
design of disclosures 107–10 subprime mortgages 120, 125–6,
steps in the right direction 110–12 136–41, 146, 146–53, 156–8, 179
economic significance 60–3 delinquency 120–1, 124, 154, 168–71
fees 66–7, 72–4, 76, 79–81, 91–2, 94–6, demand 7–10, 44, 99, 156, 160, 213,
109–12 223–4
and financial distress 98–100, 112 depository institutions 65, 95, 130–1, 134
functions 56–7 disclosure 4–5, 31–8, 40–3, 105–11, 125,
history 57–60 183–4, 238–40
and hyperbolic discounting 83–6, 89, annual percentage rates see annual
113–14 percentage rates (APRs),
interest rates 58–9, 66, 75, 78, 92–3, disclosure
95, 106 average-use 35, 241–2
long-term 70–2 cellular service see cellular service,
issuers 4, 14, 58–9, 65, 69–70, 72–3, disclosure
103–4 comprehensive 37, 43, 240, 244
market 54–66, 76–7, 82, 96–8, 100–3, electronic 41, 43, 110
105–6, 114 individual-use 35–6, 242–3
competition 64–5 mobile 246
participants 63–4 product-attribute information 35, 55,
structure 63–5 106, 192, 240, 243, 245
penalty fees 66–7, 72–4, 76, 90, 94, 100 product-use information 4, 31,
pricing 54, 66, 69–70, 81, 90, 97, 112 33–6, 55
products 10, 33–4, 53–4, 65, 68, 79, credit cards 106–7
105–6 design of disclosures 107–10
rethinking of disclosure 105–12 steps in the right
credit limits 10, 12, 64, 72–3, 76, 90, 110 direction 110–12
credit unions 102–4 real-time 110, 245
cross-subsidization 18, 75–6, 154 regulation 4, 13, 32–43, 56, 106, 174,
cumulative distribution functions 213–14 192–3
currency-conversion fees 27, 52–3, 73–4, cellular service contracts 238–40
80, 91–2, 98–9 normative questions 41–3
customer-acquisition costs 189, 207, 223 optimal disclosure mandates 36–41
product-use information 33–6
data services 1, 188, 209, 229, 234, 245 simple 4, 37, 43, 110
debit cards 56–7, 101, 103–5 timing-of-disclosure problem 126,
default, subprime mortgages 144–5, 155 180–1
default interest rates 18, 26, 52, 66, 74, total cost 112, 175, 243, 245
76, 100 use-pattern information 192, 241–3
deferred costs 4, 17, 21–6, 52–4, 77–8, discounts 21–2, 69, 82–6, 89, 91,
169, 183–5 113–14, 170
274 i ndex
distorted competition 23–5 Federal Trade Commission (FTC) 40,

and credit cards 98–9 132–4, 138, 161
and subprime mortgages 167–8 fees 18–19, 52–3, 76, 118–22, 142–4, 154,
distributional concerns 4, 7, 25–6 172–6
credit cards 55, 98, 100–1, 112 in the cell phone market
subprime mortgages 166, 172–4 fixed 211, 216–17
down payments 135–7, 146–7, 157 monthly 1, 191, 206, 208, 211–12,
small 119, 136, 168 216, 229
Driscoll, J.C. 29, 79–80, 144, 162, 165 overage 1, 27, 191–2, 213–14, 217,
233–4, 242
early termination fees (ETFs) 187, in the credit card market 66–7, 72–4,
192, 206–8, 210–11, 222–6, 76, 79–81, 91–2, 94–6, 109–12
231, 236 balance-transfer 52, 66, 95
graduated 201, 207, 210, 236 convenience 74, 91, 95
lock-in enforced by 224–5 currency-conversion 27, 52–3, 73–4,
education 7, 26, 28, 30–1, 35, 173 80, 91–2, 98–9
financial 103, 161 expedited-payment 52, 76, 95
by sellers 30–2 interchange 64, 66, 69, 93, 96, 100
efficiency 2, 15–16, 23, 43, 49, 52, 75 late payment 72, 96, 99
costs 7, 24, 231–2 no-activity 52, 66, 74
efficiency-based explanations 2, 75, 77, 154 over-limit 52, 66, 72–3, 76, 90, 94–6,
efficient risk-based pricing 18, 53, 108–9
75–6, 120 overdraft 104
electronic disclosure 41, 43, 110 risk-based 95–6
Engel, K.C. 131–2, 141, 165, 167, service 66, 74, 80, 94, 96
170, 181 in the mortgage market 142–4
equilibrium 15, 31, 50 appraisal 118, 177–8
pricing scheme 14, 48 escrow analysis 118, 143
escalating payments 118, 136–40, 148, flood certification 118, 142–3
168–9 foreclosure 119, 144, 154
escrow analysis fees 118, 143 origination 118 , 142–3 , 167 ,
ETFs see early termination fees 178–9
Evans, D. 58–65, 69, 71, 77, 83, 92–4, non-salient 97, 176
104 upfront 157, 189, 223
ex ante mistakes 215, 221–2 finance charges 27, 38, 67, 70, 72, 126,
ex post mistakes 215, 221 174
expedited-payment fees 52, 76, 95 definition 38, 127
expert advice 28–9, 165, 172 financial distress 98–100, 112
first-lien subprime loans 129–30, 137,
family plans 188, 209 140, 148
Fannie Mae 123, 137, 163, 182 fixed costs 68 , 77–8 , 189 , 200 , 216 ,
Federal Communications Commission 223
(FCC) 41, 194–7, 200–2, 222, fixed fees 211, 216–17
235, 237–9, 245–6 fixed monthly fees 1, 191, 206, 208,
Federal Deposit Insurance Corporation 211–12, 216, 229
(FDIC) 96, 133 fixed-quantity monthly
Federal Reserve Board (FRB) 65–8, packages 209–10
79–80, 125–6, 133, 147–9, fixed-rate mortgages (FRMs) 117, 137,
163–4, 169–70 140–1, 147, 153, 163, 169
i nde x 275
flood certification fees 118, 142–3 imperfect rationality 12, 20–1, 54, 101–2,
foreclosure 124, 127–9, 148, 151, 154, 166, 125–7, 174–5, 191
168–71 of borrowers 121, 124, 174–5, 180, 183
fees 119, 144, 154 of cardholders 52–4, 78–9, 82, 112
rates 116, 168–9, 171, 183 of consumers 4, 187–9, 224, 228–9,
subprime mortgages 168–71 247
FRB see Federal Reserve Board imperfect self-control 83, 86–7, 90
Freddie Mac 144, 163, 182 imperfectly informed consumers 9, 33,
free or subsidized handsets 185, 187, 189, 45
191, 207, 223–6 see also incentives 21, 30, 39, 98, 111–12, 165, 176
bundling incomes 60, 74, 83, 88, 119, 147–8,
free phones 189, 191, 207, 223, 226 172–3
frequent-flyer miles 53, 68–9, 80, 100 individual-use information 35–6, 38,
FRMs see fixed-rate mortgages 107–9, 111, 242–3, 245
FTC see Federal Trade Commission inefficient allocation 23, 25, 98, 166,
Furletti, M. 36, 60, 66–9, 71–2, 76, 95–6, 183, 230
106–7 inefficient contracts 16, 98–9, 117
intensity 11, 13, 45–6, 64
Gabaix, X. 19, 31, 79–80, 162 interchange fees 64, 66, 69, 93, 96, 100
Government Accountability Office interest payments 71, 75, 113, 138, 150
(GAO) 36, 72, 132 interest-rate risk 120, 153–4
graduated early termination fees interest rates 10–12, 33–4, 52, 153–4,
(ETFs) 192, 201, 207, 210, 236 157–8, 160–2, 173
Grubb, M. 35, 186, 212–15, 218 credit cards 58–9, 66, 70–2, 75, 78,
92–3, 95
handsets 19–20, 188, 202–3, 225 default 18, 26, 52, 66, 74, 76, 100
bundling of see bundling introductory 22, 25, 52, 81, 92,
free or subsidized 185, 187, 189, 191, 113, 138
207, 223–6 long-term 52, 66, 70–2, 81, 175
manufacturers 203–4, 223, 227, 236 market 58, 120, 149–50, 153
market 202–4 subprime mortgages 142
Haughwout, A. 169, 173–4 intermediaries 5, 37, 40–1, 43, 56, 110–11,
Herfindahl-Hirschman Index (HHI) 199 244–5
heterogeneity 189–90, 242 sophisticated 5, 37, 193, 244
in cognitive ability 160–4 internet 19, 65, 131–2, 209
of consumers 17, 35, 37, 39, 76, 155, 172 interpersonal learning 26–9, 164
HHI see Herfindahl-Hirschman Index introductory interest rates 22, 25, 52, 81,
hindered competition 3, 23 92, 113, 138
and credit cards 98 introductory periods 25, 70, 81–2, 92,
and subprime mortgages 166–7 113–15, 118–19, 138–9
house prices 63, 120–1, 123, 139, 151–3, low-rate 126, 137, 140, 149, 178
158, 182 issuers 47–9, 52–5, 64–9, 71–7, 80–2,
hybrid mortgage 117, 137–8, 140, 148–9, 93–103, 106–13
169 costs 15, 25, 48–50, 70, 91
hyperbolic discounting 83–6, 89, 113–14 credit cards 4, 14, 58–9, 65, 69–70,
72–3, 103–4
imperfect competition 132, 191
imperfect information 9, 18, 30, 78, 132, Kahneman, D. 8, 27
164, 229 Kaufman, A. 102
276 i ndex
Laibson, D. 19, 22, 29, 31, 70, 79–80, 84 McCoy, P.A. 131–2, 141, 165, 167, 170,
late fees 14–15, 47–50, 72–3, 94–6, 107–9, 177, 181
144, 178 McDonald, D. 34, 108
late payments 14–15, 21–2, 25, 33–4, mandatory arbitration clauses 74–5
46–50, 73–4, 107–9 Mann, R.J. 57, 62–3, 99, 104, 110, 114
cost of 15, 25, 49 market correction 164–5
incidence of 15, 33, 47–8 market failures 4, 7, 16, 32, 52, 112, 116
learning 7, 101, 161, 164–5, 177, behavioral 2–4, 16–17, 32, 42–3, 65,
190, 192 112, 237–8
by borrowers 164–5 market forces 2–4, 6–7, 13, 21, 26, 32,
by consumers 26–30, 188, 190–2, 218, 43–4
229, 231, 233–8 market interest rates 58, 120, 149–50, 153
interpersonal 26–9, 164 market shares 64, 131, 138, 202
lenders 120–2, 131–2, 150–1, 159–60, market solutions 7, 26–32, 35, 43, 56,
170–1, 173–8, 180–1 233–4, 236–7
libertarian paternalism 32 cell phones 191–2, 233–7
loan contracts 86, 117, 121, 125, 135–6, credit cards 101–5
140, 142 markets
loan officers 162, 179–80 cell phones see cell phones, market
loan originators 131, 133, 154, 165, 179 cellular service see cellular service,
loan products 121, 125–6, 155, 179–80 market
loans 117–19, 121–3, 128–32, 137–51, competitive 2, 15–16, 19, 31, 48–9, 165,
156–9, 161–70, 177–9 see also 215–17
mortgages; subprime consumer 5–6, 22, 33–4, 42–4, 246,
mortgages 248–9
closed-end 77, 86, 126 credit cards see credit cards, market
deferred-cost 120, 125–6, 146, 148, handsets 202–4
151, 179 prime mortgages 117, 119, 128, 146,
lock-in 114, 187, 189–90, 207, 223–5, 177, 181
231–2 subprime mortgages see subprime
behavioral-economics theory 224–7 mortgages, market
clauses 185, 187, 189, 206–7, 210, MasterCard 63–4, 75
223–7, 231–2 Mayer, C.J. 129–30, 136–41, 147, 149–50,
contracts 3, 19, 189, 191, 223, 226 153, 162, 172–4
costs of 114, 187, 235–6, 242 Meier, S. 86–7
ETF-enforced 224–5 merchant cards 57–8
rational choice theories 223–4 misperceptions 2, 8–14, 16–17, 44–8,
welfare implications 231 90–1, 185–7, 215–16
long-term affordability 147–8, 158 cardholder 55–6
long-term contracts 187, 206–7, 225 consumer 15, 17, 20, 30–1, 45,
long-term costs 1, 39, 112, 120, 158, 105–6, 186
175–6, 185 design in response to 8
long-term interest rates 52, 66, 70–2, objects of misperception 10–14, 45–6
81, 175 systematic 7
long-term prices 23–4, 26, 81–2, 91–3, mistakes 28–9, 164, 190–2, 220–3,
100–1, 121, 125 229–30, 232–3, 237
low-rate introductory periods 126, 137, consumer 32, 38, 165, 189–90, 212, 230,
140, 149, 178 232–3
loyalty programs 52, 65 correction by sellers 165
i nde x 277
ex ante 215, 221–2 non-salient fees 97, 176

ex post 215, 221 non-salient price dimensions 18–19, 21,
product-use 13, 15, 29, 33, 50 25, 53, 80–1, 94, 159–60
use-pattern 13, 223 non-salient prices 15, 19, 24–6, 39, 50,
mobile disclosure 246 53, 81
monthly charges 186–7, 205, 211, 224, nontraditional mortgages 159, 168
228
monthly payments 46, 72, 90, 118, 137–9, objects of misperception 10–14, 45–6
142, 178 Office of Thrift Supervision
monthly statements 55, 109–11 (OTS) 133–4
mortgage loans see mortgages; subprime optimal contract 81, 156, 167–8
mortgages optimal disclosure design 40, 42, 108,
mortgages 1, 5–6, 22, 27, 36, 38, 116–84 110, 112
adjustable-rate see ARMs optimism 2, 21–2, 53–4, 82–3, 88–90,
behavioral-economics theory 121, 123, 95, 158
156–65 bias 22, 82, 88, 121
contracts 135–46 option adjustable-rate mortgages 138–9,
design 116, 117–23, 131, 133, 146, 169 142, 145, 148, 158–9
fixed-rate 117, 137, 140–1, 147, 153, origination fees 118, 142–3, 167, 178–9
163, 169 originators 130–1, 133, 154, 158, 181
hybrid 137, 149 OTS see Office of Thrift Supervision
nontraditional 159, 168 outstanding balances 60–1 , 72, 114 ,
option 139, 148 140
policy implications 125–7 over-limit fees 52, 66, 72–3, 76, 90, 94–6,
prepayment see prepayment 108–9
prime loans 141, 162–3, 173, 178 overage charges 191, 216–17, 221, 228,
prime market 117, 119, 128, 146, 233–5, 237, 242–3
177, 181 overage fees 1, 27, 191–2, 213–14, 217,
rational choice theories 119–21, 233–4, 242
146–56, 160 overconfidence 27, 35, 186, 189, 235
subprime see subprime mortgages overdraft fees 104
welfare implications 123–4, 166–74
multidimensional contracts 3, 97, 121, paternalism 42–3
124, 166, 183, 189 asymmetric 32
multidimensional pricing 18–19, 52–3, libertarian 32
80, 119, 125, 141, 145 soft 32
multidimensionality 18, 53, 80, 120, 183, pay-as-you-go 209, 211, 236
188–9, 227–9 payment-option adjustable-rate
myopia 21–3, 39, 53, 82, 121, 158, 175–6 mortgages 118, 123, 147, 158
payment shock 118, 138–9, 169
national carriers 185, 196–8, 211 peak minutes 1, 217
National Credit Union Administration penalty fees, credit cards 66–7, 72–4, 76,
(NCUA) 133 90, 94, 100
negative amortization 118, 139, 142, 157 per-transaction fees 52, 54, 68–9, 77–8,
net benefit 17, 24, 48–50, 99, 185 93, 99–100
networks 58, 196, 198, 201, 203–4, 228, per-use prices 11–14, 45–7
236 perceived total price 3, 14–15, 19, 21, 23,
Nextel 194, 196, 199 25, 50
no-activity fees 52, 66, 74 phones see cell phones; handsets
278 i ndex
post-introductory rates 70, 92, 113–14, imperfect see imperfect rationality

126, 138 real-estate boom 123–4
postpaid plans 188, 205–6, 208–11, 221, real-time disclosure 110, 245
224, 235–6 redistribution, regressive 26, 190, 230, 232
prepaid plans 188, 190, 198, 210–11, 221, refinancing 22, 29, 141, 144, 149–50, 152,
235–7 157
prepayment 119, 126–7, 144–5, 150, 155, regressive redistribution 26, 190, 230, 232
178–80, 182–4 regulatory strategies 33, 106, 248–9 see
penalties 139–41, 144, 150, 157, 163, also disclosure, regulation;
168–9, 178–9 subprime mortgages, market,
prices regulatory scheme
house 63, 120–1, 123, 139, 151–3, 158, Renuart, E. 109, 111, 126–7, 141–3,
182 174–6, 178–9, 181
long-term 23–4, 26, 81–2, 91–3, 100–1, repayment 38, 82, 89–90, 111
121, 125 residential mortgage market 117, 127,
non-salient 15, 19, 24–6, 39, 50, 53, 81 130, 138, 160
pricing revenues 9–10, 44, 49, 69, 72–5, 96,
credit cards 54, 66, 69–70, 81, 90, 97, 112 229–30
multidimensional 19, 53, 80, 119, 141, total 72–3, 96, 204
145, 159 risk aversion 214–15
prime loans 141, 162–3, 173, 178 risk-based fees 95–6
prime market 117, 119, 128, 146, 177, 181 risk-based pricing 18, 60, 76, 95, 105,
product-attribute information 4, 13, 154, 177
108–11, 192–3, 240–1, 243–4, efficient 18, 53, 75–6, 120
246–7 roaming charges 206, 211, 239
disclosure 35, 55, 106, 192, 240, 243, rollover minutes 188, 228, 234–5
245
product-use information 4–5, 13, 28, 31, salience 21, 24–5, 78, 99
34–5, 55–6, 106–10 and behavioral-economics
disclosure 4, 31, 55, 106, 110 theory 91–6
regulation 33–6 and contract design 92–4
product-use mistakes 13, 15, 29, 33, 50 and credit cards 91–6
profits 8–9, 31, 45, 97, 102, 197, 232 dynamics of 94–6
psychology 7–9, 11, 13, 15, 17, 19, 27 salient price dimensions 23–4, 94, 97,
consumer 2–4, 6–8, 15, 17, 26, 32, 43–4 124, 159, 166, 191
salient prices 3, 16, 24–5, 50, 96, 99, 188
rational-choice models/theories/ securitization 116, 120, 122, 124, 129,
explanations 3, 8–9, 14, 188–9, 131–3, 137
212–13, 223, 227 securitizers 130, 133, 154
and cell phones 188–9 self-control 54, 70, 82, 84, 86, 89–90
and cellular service contracts 212–15, problems 54, 82, 86–7
223–4, 227 sellers 1–4, 7–10, 16–17, 23–6, 28, 30–2,
and credit cards 75–8 37–44
and deferred costs 77–8, 146–53 education by 30–2
and subprime mortgages 119–21, service fees 66, 74, 80, 94, 96
146–56, 160 Shiller, R.J. 152–3, 168
rational consumers 2–5, 9–10, 16–21, short-term affordability 146–7, 150, 158
23–5, 47–9, 187–9, 227–9 short-term benefits 1, 20, 53, 82, 175, 226
rationality Shui, H. 65, 92–3, 114
bounded 2–3, 36 simple disclosures 4, 37, 43, 110
i nde x 279
social welfare see welfare competition 131–3

soft paternalism 32 participants 130–1
sophisticated consumers 14–15, 30, 32, regulatory scheme 133–5
101–2, 105, 233, 237 structure 130–3
speculation 120 policy implications 174–83
mortgages 150–4 prepayment see prepayment
speculators 120, 150–3 rational choice theories 119–21,
Spiegler, R. 6–8, 17, 19, 23, 32, 115, 212 146–56, 160
Sprint 185, 195–7, 199, 205, 207–9, 228, 231 deferred costs 146–53
Stango,V. 90, 114, 176 securitized 116, 129, 137
subprime mortgages 116, 120–2, 127–9, speculation 150–4
135–6, 153–4, 156, 163–6 welfare implications 166–74
affordability 146–50 Sullivan, T.A. 51, 87, 89
and APR disclosure 174–83 Sunstein, C.R. 32, 37–8, 40–1, 43
behavioral-economics theory 121, 123, switching costs 64–5, 113–15, 201, 222
156–65 systematic misperceptions 7
complexity 158–60
deferred costs 156–8 T-Mobile 185, 194, 196–8, 205, 207–9,
heterogeneity in cognitive 228, 233
ability 160–4 tariffs
market correction 164–5 three-part 185–9, 205–6, 208, 212–13,
borrowers 118–20, 128, 130, 132, 141, 215–18, 230, 232–3
161, 163 two-part 212, 215–17, 234
contracts 135–46 tax certification 118, 142–3, 159
complexity 141–5 TBO (total-benefit-of-ownership) 38–9
deferred costs 120, 125–6, 136–41, TCO (total-cost-of-ownership) 37–9,
146, 146–53, 156–8, 179 193, 244, 247
escalating payments 118, 136–40, teaser rates 54, 69–70, 92, 94–5, 112–15,
148, 168–9 137–8, 147
prepayment penalties 139–41, 144, technology, cell phones 194–5
150, 157, 163, 168–9, 178–9 termination fees see early termination
small down payments and high fees
LTVs 136–7 Thaler, R.H. 18, 27, 32, 41, 83–4, 164
crisis 116, 122–3, 125–6, 131, 135, 146, three-part tariffs 185–9, 205–6, 208,
149 212–13, 215–18, 230, 232–3
default 144–5, 155 behavioral-economics theory 215–23
definition 128–9 rational choice theories 212–15
delinquency 120–1, 124, 154, 168–71 welfare implications 230
and disclosure of total cost of timing-of-disclosure problem 126,
credit 174–83 180–1
and distorted competition 167–8 total-benefit-of-ownership see TBO
distributional concerns 166, 172–4 total cost disclosures 112, 175, 243, 245
fees 142–4 total cost of credit 112, 125–6, 174–83
first-lien 129–30, 137, 140, 148 total-cost-of-ownership see TCO
foreclosure 168–71 total price 10–15, 19, 24–5, 34, 45–7,
and hindered competition 166–7 49–50, 193
interest rates 142 actual 3, 23, 25, 112, 220
lenders 132, 143, 147 perceived 3,14–15, 19, 21, 23, 25, 50
market 116–19, 123–4, 126, 128–36, Tracy, J. 173–4
139, 146, 148 tradeoffs 9–10, 40–1, 241
280 i ndex
transaction costs 24, 77, 144, 152, 170–1 weekend minutes 1, 211, 228, 239,
two-part tariffs 212, 215–17, 234 241, 243
welfare 3, 7, 10, 13, 19, 23–6, 32
underestimated use 50, 191, 235, 237 and cell phones 189–91
underestimation biases 54, 91, 113 and cellular service contracts 190,
unlimited calling plans 191, 208–9, 228, 229–32
233–5, 237, 243 and complexity 190, 232
upfront fees 157, 189, 223 costs 23, 54–6, 101, 166–70, 189–92,
usage 186, 206, 212–13, 215–16, 218, 230–2, 237–8
222–3, 225 and credit cards 54–5, 97–101
use-pattern information 31, 192–3, and lock-in clauses 231
240–5, 247 and mortgages 123–4, 166–74
disclosure 192, 241–3 and three-part tariffs 230
use-pattern mistakes 13, 223 Westbrook, J.L. 51, 57–63, 72, 86–9
use patterns 4–5, 10–11, 40–1, 106–7, wireless communications 185, 194–5,
110–11, 193, 223–4 200, 206
usury ceilings 58–9 wireless services 185, 187, 198, 207,
220, 226
Verizon 185, 195, 197–8, 201, 203, 205, women 121, 172–3
207–11
Visa 63–5, 75 Zinman, J. 90, 103, 176
Zywicki, T.J. 63, 77, 141, 146, 151,
Warren, E. 30, 51, 57–63, 72, 86–9, 134 172, 178

Seduction by Contract

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Seduction by Contract

Uploaded by

Copyright:

Available Formats

SEDUCTION BY CONTRACT

This book is the culmination of several years of research and writing on

This book is the culmination of several years of research and writing on

W e are all consumers. As consumers we routinely enter into contracts

Seduction by Contract. Oren Bar-Gill.

Market Forces and Consumer Psychology

within a rational-choice framework proves the robustness of the biases and

Social Costs of the Behavioral Market Failure

Finally, complexity and deferred costs raise distributional concerns, since

Toward More Effective Disclosure Mandates

information. The annual cost of cellular service would combine rate

Seduction by Contract. Oren Bar-Gill.

prices. The goal is to highlight predictions of the behavioral-economics

I. The Behavioral Economics of Consumer

A. Designing Contracts for Biased Consumers

of benefits and prices: D(B,P). The seller’s revenue is determined by the

2. The Objects of Misperception: Product Attributes and Use Patterns

for example, is used frequently by some consumers and less frequently by

The second category of objects of misperception includes use levels. Dif-

The answer depends on consumer psychology and market structure. For

salient prices are reduced. A price is non-salient, because consumers think

4. The Limits of Competition

B. Common Design Features

he approximates, rather than calculates, the impact of the salient dimensions

and a less complete understanding of the contract. Imperfect rationality,

II. Welfare Implications

outweighed by the marginal cost of collecting more information.When the

Excessive complexity imposes a larger burden on less sophisticated cus-

III. Market Solutions and Their Limits

19. See also Armstrong, above note 3.

The distinction between product-attribute information and product-

the market.23 These experts include private consultants and advisers,

Finally, even apart from the collective-action and product-use informa-

IV. Policy Implications: Disclosure Regulation

question by Omri Ben-Shahar and Carl Schneider.32 In the consumer mar-

A. Disclosing Product-Use Information

The cellular service market provides another example. A pricing manager at

Individual-use disclosures, while more effective, are also subject to cer-

B. Designing Optimal Disclosure Mandates

To be effective, disclosure mandates must adopt one of two general

1. Simple Disclosures for Consumers

actual and perceived cost of their products.The TCO disclosure undermines

ual-use information. When designing such disclosures, regulators should be

2. Comprehensive Disclosures for Intermediaries and Sellers

A related model skips the intermediary and relies on competing sellers as

51. See also Sunstein, above note 38, at 4.

consumer psychology and market forces, while following common patterns,

2. The Objects of Misperception

Table 1.1. Benefit from paying late

Table 1.2. Total benefit from late payments

Table 1.1a(1). Perceived benefit from paying late—Case (1)

Table 1.1a(2). Perceived benefit from paying late—Case (2)

Seduction by Contract. Oren Bar-Gill.

B. Contract Design Explained

insufficient attention to long-term costs. Optimistic cardholders underesti-

pricing creates an appearance of a cheaper product without actually offering

3. Pub. L. 111–203, Title X, Sec. 1033.

I conclude this chapter with guidelines for additional steps. At a general

I. The Credit Card Market

The result: An explosion of consumer credit, which led to a dramatic

Sufi, and by Michelle White, suggest a causal relationship between credit

24. Ausubel, above note 11, at 51.

II. The Credit Card Contract

pricing.30 Then we’ll explore competing rational-choice and behavioral-